epubs.surrey.ac.ukepubs.surrey.ac.uk/812312/1/e thesis 05.09.16.doc · web viewfield, a. (2013)....
TRANSCRIPT
The Relationship Between Self-Esteem, Narcissism, Psychopathy and
Aggression in a High Secure Psychiatric Population
Carly Samson
Submitted for the Degree of
Doctor of Psychology(Clinical Psychology)
School of PsychologyFaculty of Health and Medical Sciences
University of SurreyGuildford, SurreyUnited KingdomSeptember 2016
Abstract of MRP Empirical Paper
The purpose of this study is to explore the relationship between self-esteem, narcissism,
psychopathy and aggression to aid our understanding of personality factors that precipitate
aggression. Further research into factors associated with aggression is urgently required due
to the huge economic and personal costs of this behaviour. This is the first time these
relationships have been explored in a unique and hard to treat population who are at high risk
of harming others. Fifty inpatients of a high secure psychiatric hospital completed self-report
measures of personality traits and completed a computerised task measuring implicit self-
esteem. Information regarding previous and institutional aggression was obtained from
patients’ medical files. Self-esteem fragility did not predict high levels of physically
aggressive behaviour. However the findings suggest that narcissism and explicit self-esteem
may play a role in institutional aggression. A profile of high explicit and high implicit self-
esteem was found for Factor 1 psychopathy, and high explicit self-esteem was associated
with adaptive narcissism. These findings suggest that narcissism and psychopathy have
different self-esteem profiles. Alternative measures of implicit self-esteem in high secure
populations are required to further test hypotheses relating to self-esteem fragility and
aggression.
Acknowledgements
Firstly my sincere thanks to all participants who participated in the research and met with me
to offer feedback and advice regarding the design and methodology. As well as giving up
their time, many also gave me very helpful information about the conceptualisations and
ways of measuring aggression. Thank you also to all the individuals who have supported my
development and learning throughout training, as well as all the incredible people who have
shared their experiences with me and taught me so much about distress and resilience.
I would like to say a massive thank you to my supervisors, Erica Hepper and Simon Draycott,
as well as my clinical tutor Nan Holmes, who have been an invaluable source of guidance,
support and encouragement throughout the entire training process. A huge thank you also to
the staff at the University of Surrey and to my fellow trainees from whom I have learnt so
much and shared so many amazing experiences.
Throughout my life I have been fortunate to have family, friends and an amazing husband
who have kept me grounded, reminded me of the importance of a good work/life balance and
encouraged me to achieve my goals and dreams. Thank you for laughing and crying with me,
and for always believing in me.
Contents
Page Section Title
1 Research MRP Empirical Paper
76 MRP Empirical Paper Appendices
206 MRP Proposal (without appendices)
232 MRP Literature Review (with appendices if appropriate)
296 Clinical Experience
Details of clinical experience, including the nature of each placement and the range of clinical experience gained on each
299 Assessments Details of academic assessments
The Relationship Between Self-Esteem, Narcissism, Psychopathy and Aggression in a High Secure Psychiatric Population
By
Carly Samson
Submitted in partial fulfilment of the degree of Doctor of Psychology (Clinical Psychology)
School of Psychology
Faculty of Arts and Human Sciences
University of Surrey
March 2016
Word Count: 9884 (excluding the statement of journal choice, contents page, abstract, words within tables and figures and references and
appendices)
© Carly Samson 2016
1
Statement of Journal Choice
I have chosen to submit this empirical paper to “The Journal of Personality
Disorders”. I believe this study fits the scope of the journal (see Appendix
1), which publishes articles relating methods of understanding and treating
personality disorders. I believe that, by publishing this paper in this
international journal, the findings would reach a range of professionals
working in a variety of settings, who may find the conclusions and
implications useful in their clinical practice.
2
Abstract
The purpose of this study is to explore the relationship between self-esteem,
narcissism, psychopathy and aggression to aid our understanding of
personality factors that precipitate aggression. Further research into factors
associated with aggression is urgently required due to the huge economic
and personal cost of this behaviour. This is the first time these relationships
have been explored in a unique and hard to treat population who are at high
risk of harming others. Fifty inpatients of a high secure psychiatric hospital
completed self-report measures of personality traits and completed a
computerised task measuring implicit self-esteem. Information regarding
previous and institutional aggression was obtained from patients’ medical
files. Self-esteem fragility did not predict high levels of physically
aggressive behaviour. However the findings suggest that narcissism and
explicit self-esteem may play a role in institutional aggression. A profile of
high explicit and high implicit self-esteem was found for Factor 1
psychopathy, and high explicit self-esteem was associated with adaptive
narcissism. These findings suggest that narcissism and psychopathy have
different self-esteem profiles. Alternative measures of implicit self-esteem
in high secure populations are required to further test hypotheses relating to
self-esteem fragility and aggression.
3
Introduction
Violence and aggression present significant economic and social problems
for society, individuals, health and forensic services, causing lasting
physical and emotional damage (Short et al. 2012). The Institute for
Economics & Peace (2013) reported that violence costs the British economy
£124 billion per year. On an individual level, victims of violence are
evidently at risk of physical injury and potentially loss of life in serious
cases. Longer term, it has been demonstrated that exposure to violence is
significantly related to psychological distress (Singer, et al. 1995).
Aggression refers to behaviour carried out with the intention to harm or gain
advantage over others, towards people who are motivated to avoid harm
(DeWall & Anderson, 2011). There are many ways of measuring aggression
and methods used in NHS and forensic services vary considerably, typically
including the number of convictions, recidivism rates and frequency of
aggressive incidents (NHS Protect, 2011; NHS England, 2014). Violence
comprises the use of physical force or power against a victim and usually
refers to the most severe types of physical aggression (Howells & Hollin,
1989; DeWall & Anderson, 2011). Statutory definitions of violent crime in
the UK include physical assault with and without injury, sexual offences
and robbery (Office for National Statistics, 2012). The focus of the present
study is on direct, physical aggression directed outwards, towards other
people or property in line with the definitions above. Exploration of
individual difference factors that predispose people to aggression is a vital
part of developing our understanding of this behaviour and subsequently
how it can be managed effectively.
4
Aggression is likely caused by a number of factors which overlap and
interact. There has been a large amount of research exploring the possible
causes and moderators of aggression, including individual and
environmental factors. Life experiences such exposure to violence in
childhood (Lansford et al. 2002) and trauma (Sarchiapone et al., 2009) have
been proposed as a risk factors for aggression later in life (Farrington,
1998), and are likely to lead to the development of low self-worth (Guerra,
Huesmann & Spindler, 2003; Young, Klosko & Weishaar, 2006). People
who have a core view of themselves as defective may be hypersensitive to
perceived threats to their ego and may compensate by behaving in an
aggressive way when threatened (Baumeister, Smart & Boden, 1996;
Baumeister, Bushman & Campbell, 2000). Thus difficult life experiences
are likely to lead to negative feelings of self-worth, which may need to be
defended against in order for the individual to function. Some of these
defences may include aggressive behaviour.
Self-Esteem
The self is fundamental to how we experience ourselves, others and the
world around us, therefore understanding the various aspects of the self is
vital to the scientific understanding of personality and behaviour. The role
of the self in social and psychological functioning is well documented
(Campbell, Assanand & Di Paula, 2000), and self-disturbances are
prominent features of many mental health difficulties (Edwards et al., 2012;
Fonagy, 1999; Bateman & Fonagy, 2004; Zeigler-Hill, 2011; APA, 2013).
The self is made up of many facets, including self-esteem, self-concept,
identity and self-efficacy. Disturbances in self-esteem in particular have
5
been highlighted as important risk factors in aggressive behaviour
(Baumeister et al., 1996; Donnellan et al., 2005; Kantor & Jasinski, 1998).
Self-esteem is an evaluation of worth; the extent to which one sees oneself
as competent and worthwhile (Leary & Tangney, 2005). Until recently self-
esteem has been understood in a uni-lateral way; as either a level of positive
or negative attitude towards oneself (Sedikides & Gregg, 2003). However
dual processing models highlight that information can be processed
implicitly and explicitly simultaneously, with the former being outside of
conscious awareness (Greenwald & Banaji, 1995). The Full Discrepancy
Model (Kernis, 2005) links these models to self-esteem, proposing that there
are individual differences in conscious (explicit) and unconscious (implicit)
representations of the self. Congruence of high explicit and implicit levels
are thought to represent secure self-esteem whereas a discrepancy between
them represents fragile/insecure self-esteem. Failure to capture the various
dimensions of self-esteem has limited our understanding of this construct in
the current evidence base.
The security of self-esteem plays an important role in psychological
wellbeing and the ability to cope with adverse events (Kernis, 2003). For
example people with high, secure self-esteem can draw upon internalised
feelings of worth that can be used to maintain positive feelings when
negative events occur (Steele and Aronson, 1995), whereas those who have
fragile self-esteem are less able to maintain such balance (Malle &
Horowitz, 1995). There are two key models of fragile self-esteem; the Full
Discrepancy Model and the Partial-Discrepancy Model (Gregg & Sedikides,
2010). The Full Discrepancy Model defines fragile self-esteem as a
6
discrepancy between ESE and ISE, with larger discrepancies indicative of
more fragile SE, however it does not specify the nature of the discrepancy.
The Partial Discrepancy Model goes further by stating that fragile self-
esteem constitutes specifically high ESE and low ISE, operationalised as a
significant interaction between ISE and ESE. Follow up simple slopes tests
exploring the direction of the interaction would estimate whether it was
fragile self-esteem (high ESE and low ISE) or some other combination (e.g.
high congruent or low congruent) that was associated with the highest level
of their outcome variables (Schroder-Abe et al., 2007; Gregg and Sedikides,
2010). This particular self-esteem combination has been associated with
defensive, aggressive reactions following an ego threat (Jordan et al., 2003).
Three key reviews of the evidence of discrepant implicit and explicit self-
esteem in relation to aggression have been examined in non-clinical samples
which highlight the inconsistent findings regarding the association between
these concepts (Walker & Bright, 2009a; Ostrowsky, 2010; Walker &
Knauer, 2011). The focus on global self-esteem, as well as problems
developing reliable and valid measures of the various dimensions of self-
esteem are likely to explain the mixed findings in this area. Ostrowsky
(2010) suggested that high and low self-esteem may be related to different
types of aggression, such as instrumental and reactive aggression
respectively. Indeed, many studies of the relationship between self-esteem
and aggression have used measures which do not differentiate between
types of aggression such as self-report measures of aggressive behaviour
(Gillespie, 2005; Falkenbach, Howe & Falki, 2013), ratings of aggression
(Goldberg et al., 2007) and informant reports (Cale & Lilienfeld, 2006).
7
These methodological issues need to be addressed in order to explore the
relationship between self-esteem and aggression in more depth.
Self-esteem fragility is a central component of many psychiatric syndromes
such as Narcissistic Personality Disorder (NPD) and Psychopathy (APA,
2013; Cleckley, 1941; Hare, 2003), both of which are consistently
associated with aggression in research (for reviews see Reidy, Shelley-
Tremblay & Lilienfeld, 2011; Coid & Roberts, 2007 & Baumeister et al,
2000). It is therefore surprising that the ideas developed in non-clinical
samples have not yet been tested in clinical and forensic populations where
the risk of aggression towards others is high. Most of the research with
clinical samples have used uni-dimensional measures of self-esteem and are
only beginning to consider the additional influence of other factors such as
narcissism and psychopathy (Cale & Lilienfeld, 2006; Svindseth et al.,
2008; Gillespie, 2005). A small number of studies have found that
discrepancies between explicit self-esteem (ESE) and implicit self-esteem
(ISE) are associated with higher levels of psychiatric symptoms (Schroder-
Abe et al, 2007; Vater et al., 2013), however they did not test the influence
of self-esteem discrepancies in relation to aggression.
Research with clinical and forensic samples needs to catch up with
methodological developments in self-esteem and aggression research from
studies with non-clinical populations so findings may be more usefully
applied to assessment, treatment and risk management. It may be pertinent
to study these ideas with clinical presentations thought to have high levels
of self-disturbance such as individuals with personality disorders and anti-
social, psychopathic personalities.
8
Psychopathy
Psychopathy includes personality traits such as a lack of empathy or
remorse, impulsivity, manipulativeness and is often associated with a
pervasive pattern of involvement in criminal behaviour (Cleckley, 1941;
Hare, 1991) and a heightened risk of aggression (Blackburn & Coid, 1998;
Herve, Hayes & Hare, 2003; Hare & Jutai, 1983; Hare & McPherson, 1984;
Toch, et al., 1989). It has specifically been associated with instrumental
aggression (Cornell et al., 1996; Woodworth & Porter, 2002), which
involves action to obtain a goal such as money or sexual gratification, and
planning. The objective of instrumental aggression is to obtain a goal
beyond harming the victim.
These conceptualisations carry the assumption that psychopathy is an
undesirable trait associated with social deviance, however exploration of
this construct in depth has revealed the presence of two dimensions, or
factors, of psychopathy which appear to be associated with different
behaviours and traits. Factor 1 comprises the affective and interpersonal
aspects of psychopathy (such as grandiosity and a lack of empathy or
remorse) and is negatively associated with psychological distress (Harpur,
Hare & Hakstain, 1989), and positively associated with narcissism (Miller et
al., 2008). Factor 2 is positively related to socially deviant traits and
behaviours including aggression and criminal behaviour (Hare, 1991),
recidivism (Hemphill, Hare & Wong, 1998), impulsivity (Miller et al.,
2008), as well as high levels of distress (Verona, Patrick, & Joiner, 2001).
As Factor 1 and 2 psychopathy appear to have different relationships with
aggression, narcissism and psychological distress, they may also have
9
different self-esteem profiles which could explain the risk of, and function
of, aggression for these individuals. It may also allow us to explore whether
psychopathy really is a maladaptive trait.
Research into ESE and ISE of psychopathic individuals is surprisingly
sparse. Pathological aspects of psychopathy have been linked to a fragile
perceptions of self-worth, a hypersensitivity to criticism and an inclination
towards aggression (Bushman and Baumeister, 1998; Cale & Lilienfeld,
2006; Kernberg 1975). However a closer examination of the sub-types of
psychopathy in relation to self-esteem and aggression may elucidate the
relationships between these constructs in clinical samples.
Narcissism
Narcissism is conceptualized as an inflated view of the self and a motivation
to be regarded as better than others (Campbell, 1999), whereas self-esteem
is understood as a global evaluation of self-worth. Current theoretical
understandings of the self-esteem profile of narcissism (known as “mask
models”) view grandiosely high ESE is a way of protecting underlying low
levels of ISE (Kohut 1966, 1977; Kernberg, 1975; Campbell & Foster,
2007; Gregg & Sedikides, 2010; Kernis, 2003) and posit that, in extreme
circumstances, aggressive behaviour is used to defend a vulnerable core self
from a perceived threat. While psychopathy has been associated with
instrumental aggression, narcissism has been linked to reactive aggression
(Barry et al., 2007; Baumeister et al., 2000), which involves a reaction to
perceived threat or provocation with high arousal levels. The objective of
reactive aggression is to harm the victim in response to strong feelings
aroused by the victim’s behaviour (Cornell et al., 1996).
10
Most of the empirical research into self-esteem, narcissism and aggression
comes from non-clinical populations exploring sub-clinical levels of
narcissism, with inconsistent findings (Bosson et al., 2008; Busman &
Baumeister, 1998; Falkenbach et al., 2013). In clinical samples the evidence
linking self-esteem and narcissism is also inconsistent (Pincus et al., 2009;
Svindseth et al., 2008), which may be due to differences in the measurement
of these constructs. Examination of the psychometric properties of
narcissism measures have revealed the presence of different sub-types of
narcissism which are associated with different self-esteem profiles (Rose,
2002) and behavioural outcomes (Barry et al., 2007; Ackerman, 2011).
Narcissistic traits such as exploitativeness and exhibitionism are considered
maladaptive and have been linked to poor social adjustment (Raskin &
Terry, 1988; Barry et al., 2007) and aggressive behaviour (Washburn et al.,
2004), whereas adaptive narcissistic traits such as authority and self-
sufficiency are associated with self-confidence, assertiveness and fewer
conduct problems (Raskin & Terry, 1988).
Studies with psychiatric populations support the distinction between
adaptive and maladaptive narcissism. Pathological/maladaptive narcissism
has been found to correlate negatively with ESE in psychiatric patients
(Pincus et al., 2009), whereas a positive relationship has been found
between adaptive narcissism and ESE (Bosson et al., 2008; Bushman &
Baumeister, 1998). Thus maladaptive but not adaptive, narcissism may
function as a way of regulating inner feelings of inadequacy in people
diagnosed with personality disorders (PDs), particularly Narcissistic PD
(Zeigler-Hill & Jordan, 2011). When these regulatory strategies fail,
11
individuals may resort to aggressive behaviour to protect their egos from
injury. It is therefore important to study the two sub-types of narcissism in
relation to aggression in clinical samples because they may have different
self-esteem profiles which could be related to aggression in different ways.
Current models of narcissism, self-esteem and aggression ignore the
adaptive and maladaptive sub-types of narcissism and require testing in
forensic psychiatric populations. Understanding narcissism as a trait that can
be adaptive or maladaptive could help us make sense of conflicting findings
in the evidence base and will be central to our understanding of self-esteem
in relation to aggression.
Rationale for the Present Study
Despite the developments in our understanding of the importance of
implicit, as well as explicit processes involved in self-esteem, and the
various facets of narcissism and psychopathy, research in clinical and
forensic psychiatric settings are still using uni-dimensional measures that
are failing to capture the complexity of these relationships. These ideas
require testing in samples with pathological levels of personality- and self-
dysfunction where the risk of aggression is high so findings can be more
usefully applied to clinical practice.
With this in mind the primary aim of the present study is to quantitatively
examine the relationship between self-esteem (including fragile self-esteem
conceptualised as high ESE and low ISE) and aggression. There are a
number of ways of conceptualising fragile self-esteem, for example
measuring how stable ESE is over time (Kernis, 2005) and dissonance
between ESE and ISE, specifically high ESE and low ISE (Gregg &
12
Sedikides, 2010). The method of testing significant interactions between
ISE and ESE and conducting follow up simple slopes tests to explore the
direction of the interaction (Gregg & Sedikides, 2010; Schroder-Abe et al.,
2007; Borton et al., 2012; Zeigler-Hill, 2006) was used to test whether
levels of ISE moderated the effects of ESE on aggression; specifically
whether aggression increased as ISE decreased but only when ESE was
high. This is a more valid method of capturing fragile self-esteem than
others, for example calculating discrepancies between ISE and ESE
(Creemers et al., 2012) which have a number of limitations (Edwards, 2001;
Edwards, 2002; Kaplan & Saccuzzo, 2009; Shanock et al., 2010). See
Appendix 2 for a discussion of methods of assessing fragile self-esteem and
their strengths and weaknesses. Further aims are to understand the self-
esteem profile of narcissism and psychopathy. Exploring these relationships
in a sample of people who have committed serious violence and currently
pose a high risk to others may generate findings with higher ecological
validity and therefore implications will be more easily translated into
clinical risk assessment and treatment. As aggression is measured in many
different ways in the literature, various measures of this behaviour will be
used in order to capture different presentations. The hypotheses for the
study are as follows:
1. Individuals with fragile self-esteem (high explicit, low implicit) will
have higher aggression ratings (higher numbers of physically aggressive
convictions, higher severity ratings of index offence and greater number
of aggressive incidents)
2. Psychopathy will predict higher levels of aggression
13
3. Psychopathy will predict higher levels of instrumental (versus
reactive) aggression in the index offence
4. Individuals with fragile self-esteem will score higher on a measure of
psychopathy
5. Individuals with fragile self-esteem will score higher on a measure of
narcissism
Method
Participants
A total purposive sample of 50 male forensic psychiatric patients was
recruited from a high security psychiatric hospital in England using the
criteria outlined in Table 1. Eligibility was assessed by discussing with
clients’ clinical team and confirmed by file review. Active grandiose
delusions were an exclusion criteria because they are thought to be a
defence serving the function of maintaining self-esteem (Bentall, 1994).
Therefore delusions may reflect a temporary change in perception of self
rather than the nature of evaluations of self-worth one would expect to see
in people with high levels of psychopathic and narcissistic traits. Across all
wards, 108 patients were identified as suitable and 84 individuals consented
to be approached (77.7%). After discussion about the study 50 of these
patients consented to participate in this study (59.52%).
Table 1Inclusion and exclusion criteriaInclusion criteria Exclusion criteria
Fluent in English Sufficiently mentally stable
to participate and able to give informed consent (this was checked with the nurse in charge on the day)
Violent forensic history
Developmental disorder or traumatic brain injury
Active grandiose delusions (i.e. fixed beliefs)
14
Previous studies looking at the effect of self-esteem on aggressive behaviour
presented effect sizes of r=-0.31 in a forensic sample (Fruehwald et al.,
1998) and r=-0.36 in an undergraduate sample (Falkenbach et al, 2013).
Assuming a power of 0.8 to detect an effect size of 0.335 (average of the
two effect sizes), one-tailed with alpha=0.05 using a correlational test, an a
priori calculation using G*Power 3.1.7 (Faul, Erdfelder, Lang & Buchner,
2007) suggested a sample size of 51 participants was needed.
Demographic information for the sample is presented in Table 2. The
majority of participants (n=27) had two or more diagnoses recorded in their
medical notes. The most common diagnosis was Anti-Social PD (n=24).
Table 2Demographic informationParameter NAge (years)
Mean (SD) 39.34 (8.97)
EthnicityWhite British 29White Other 3
Black African 1Black Caribbean 8
Black Other 1Mixed 8
Treatment PathwayPD diagnosis 18MI diagnosis 13
Mixed PD/MI diagnosis 19
Type of WardHigh Dependency
Medium DependencyRehabilitation
Admission
17402
PD=Personality disorder, MI= mental illness (refers to Axis 1 diagnosis). Patients seen under the MI treatment pathway receive treatment targeted towards their mental health symptoms, whereas the PD pathway constitutes treatments for personality disorders.
15
Measures
Explicit Self-Esteem (ESE)
The Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) is of 10 self-
report questionnaire measuring global self-esteem with a maximum score of
30. Items are rated on a 4-point Likert scale (0= strongly disagree to 3=
strongly agree), with higher score indicating higher self-esteem. An example
item is “I feel that I have a number of good qualities”. The scale has been
validated for use with a variety of clinical groups, with satisfactory internal
consistency (Cronbach’s alpha =.78) and validity reported in a prison
sample (Cale & Lilienfeld, 2006). This instrument is the most widely used
measure of ESE and allows comparison of this study with those from other
studies in this area. Good internal consistency was found in the present
study (Cronbach’s alpha=. 89).
Implicit Self-Esteem (ISE)
The Implicit Association Test (IAT, Greenwald, McGhee & Schwartz,
1998) was used to measure ISE. It is a computerised reaction time task
which measures the strength of unconscious associations between self-
relevant and non-self-relevant words with pleasant and unpleasant words
across five blocks. Blocks were presented in the same order for each
participant (1-5). The IAT used in this study was created using the FreeIAT
software (Meade, 2009, see Appendix 3 for full details of the IAT used in
this study) and run on an NHS laptop meeting the necessary security
requirements of the hospital.
16
The IAT is the most widely used measure of ISE and it has shown the
highest reliability of all the available measures of ISE (Bosson et al, 2000;
Krause et al. 2011). The list of pleasant and unpleasant words used in the
present IAT was adapted from Gregg and Sedikides (2010).
To check whether these stimuli was appropriate for this patient group,
meetings were arranged with four patient representatives from the hospital
to check whether the unpleasant words from the IAT may cause distress.
Following feedback, two words were replaced with words of similar
valance, arousal and dominance using words from The Affective Norms for
English Words (ANEW; Bradley & Lang, 2010, see Appendix 3). Patient
representatives believed that the word “bomb” would be particularly
distressing for patients experiencing paranoia and/or from a Muslim
background, and some may strongly associate “self” words to “murder” due
to their forensic history rather than because of low-implicit self-esteem. The
revised list of words used in the IAT in this study can be found in Table 3.
Table 3List of words used in the present IATSelf Non-Self Pleasant UnpleasantMeMyselfI
TheyThemThose
ExcellentHeavenJoyTrustPeaceEnjoymentFriendHonestSweetheartLoveFreedomParadise
TerrifiedCancerWarDisasterHatredSlaughterStressAgonyTortureSlimeFilthTraitor
Note: Words adapted from Gregg and Sedikides (2010) with “bomb” and “murder” replaced by “stress” and “terrified”
17
T-tests showed that the valence, arousal, and dominance scores for the
original word list (M=4.55, SD=.94) were not significantly different from
the revised word list (M=4.39, SD=.91, t(22)=.42, p=.68). See Appendix 4
for full details of the t-test analyses. In the current study, the IAT appeared
to have poor internal consistency (split half reliability, r=.54).
Although there are a number of methods of scoring the IAT (see Appendix
5), an evaluation of a number of scoring procedures concluded that the D
algorithm, which calculates the difference in mean latencies between block
5 and block 3 divided by the standard deviation of trails within these
respective blocks, is the best performing calculation (Greenwald, Nosek &
Banaji, 2003). Therefore this method was used for calculating the IAT score
in this study with one modification: latencies below 150ms and above 5000
ms were re-coded to those boundary values as there were some extreme
latencies in the data set, which may have substantially affected the overall
score (Gregg & Sedikides 2010, see Appendix 4). Negative values of d
signify low implicit self-esteem whereas positive values indicate high ISE.
Psychopathy
Self-Report Psychopathy
The Levenson Self-Report Psychopathy Scale (SRPS; Levenson et al, 1995)
is a 26 item self-report measure which is designed to assess personality and
behavioural traits associated with psychopathy. Participants were required to
rate themselves on 26 items according to a four point scale (1=strongly
disagree to 4=strongly agree). Higher scores are associated with higher
levels of psychopathy. This measure constitutes two factors; Primary
Psychopathy (Factor 1) which measures a manipulative, callous
18
interpersonal style and Secondary Psychopathy (Factor 2) which assesses
behavioural aspects of psychopathy such as inability to learn from mistakes
and impulsivity. An example item from Factor 1 is “For me, what's right is
whatever I can get away with” and Factor 2 is “I don't plan anything very far
in advance”. These two factors correspond approximately to the two factors
of the Hare Psychopathy Checklist (PCL-R; Harpur, Hare & Hakistan,
1989). The SRPS was chosen to measure self-report psychopathy because it
is the least time consuming for participants compared to other lengthy
scales, yet retains good convergent and discriminative validity (Sellbom,
2011). Good internal consistency was found in the present study
(Cronbach’s alpha= .83).
Clinician-Rated Psychopathy
The PCL-R (Psychopathy Checklist Revised; Hare, 1991) is a 20 item
checklist which can be rated by trained clinicians from an interview and
reviewing patient records. The PCL-R generates a Factor 1 (self-
centeredness and exploitation of others) and Factor 2 score (impulsivity and
anti-social lifestyle), as well as a total score. Hare (1991) reports interclass
correlations of .90 to .95 for a forensic psychiatric sample. Where PCL-R
scores were not available in patients’ notes (n=13), a clinical psychologist
trained to administer the PCL-R rated the checklist by file review. Evidence
demonstrates that the PCL-R can be rated adequately in the absence of an
interview (Williamson et al., 1987; Wong, 1988). Internal consistency could
not be calculated for the present study as individual items scores were not
available in patients’ notes.
Narcissism
19
The Narcissistic Personality Inventory (NPI) is self-report scale designed to
assess narcissistic traits consisting of 40 items in which participants are
asked to choose one of two statements which best represents them (Raskin
& Hall, 1979). For example: (A) I can usually talk my way out of anything
or (B) I try to accept the consequences of my behaviour. This scale is
widely used in social science and personality research and has good internal
consistency in forensic samples (Cale & Lillenfeld, 2006). NPI scores
correlate significantly with observer and self-report measures of narcissism
(Raskin & Terry, 1988), indicating satisfactory validity. Good internal
consistency was found in the present study (Cronbach’s alpha=.87).
Aggression
Physical aggression directed outwards, towards others or towards property
was measured in three ways:
The number of previous physically aggressive convictions which was
counted using information from participants’ medical files. Offences
considered aggressive were behaviours that involved physical violence
towards others such as Actual Bodily Harm (ABH), Grievous Bodily Harm
(GBH), battery, assault, rape and included robbery which is classified as an
aggressive act in UK definitions of aggressive offences (Office for National
Statistics, 2015).
The number of physically violent incidents on the ward over the past 12
months from date of participation as recorded on the hospital incident
reporting system was counted. Incidents were counted as aggressive if the
participant had perpetrated an act of physical aggression towards another or
20
damage to property (for example throwing objects, hitting others and
smashing furniture).
The Modified Overt Aggression Scale (MOAS; Kay, Wolkenfeld & Murrill,
1988) was used to rate the severity of aggression involved in the index
offence. The MOAS assesses four dimensions of aggressive behaviour;
verbal aggression, aggression against property, auto-aggression and physical
aggression. Each dimension required the rater to rate one of four options
from 0 (not present) to 4 indicating high severity, with high scores
indicating more aggression. Dimension scores are weighted as follows:
verbal aggression is multiplied by 1, aggression against property multiplied
by 2, auto-aggression multiplied by 3 and physical aggression multiplied by
4. Weighted scores are then summed to give a total score ranging from 0-40.
Kay et al. (1988) reported good inter-rater reliability (r=0.94), and good
discriminative validity for distinguishing high, intermediate and low violent
groups.
Ten percent each of scored MOAS forms for index offence were randomly
selected and passed to a clinical psychologist to rate to establish the degree
of inter-rater reliability, which was very good in the present study (r=.97).
Type of Aggression
The Coding Guide for Violent Incidents (CGVI; Cornell et al., 1996) was
used to rate the extent to which index offences involved instrumental or
reactive aggression. The coding guide was used to rate the presence of
instrumental and reactive aggression in the index offence on a four point
scale (1=clearly instrumental aggression, 2=primarily instrumental, some
reactive qualities, 3=primarily reactive, some instrumental qualities,
21
4=clearly reactive hostile aggression). The CGVI has been used in forensic
psychiatric settings (Laurell, Belfrage & Hellstrom, 2010) and has
satisfactory inter-rater reliability (kappa coefficient of .85; Cornell et al.
1996). Very good inter-rater reliability was found in the present study
(r=.95).
Ethical Issues
Feedback from patient representatives was incorporated in the design of the
study and related documents and measures. Permission to conduct this study
was given by the hospital forensic clinical research panel. This study was
reviewed and received a Favourable Ethical Opinion (FEO) from a National
Research Ethics Committee, the mental health NHS trust research and
development department, and the University of Surrey Faculty of Arts and
Human Sciences Ethics Committee (see Appendix 6).
Procedure
A cross-sectional, correlational design was used. During team meetings,
staff identified patients who were eligible to participate in the present study.
A staff member then asked suitable patients whether they wished to be
approached for research purposes. Patients who consented to be approached
were provided with an information sheet and given the opportunity to ask
questions with myself. Patients were given at least 24 hours to decide
whether they wished to participate. Patients who decided to participate were
invited to discuss the study in a quiet room on their ward. Patients were
asked to sign a consent form then and completed the computerised IAT,
followed by the questionnaires which were counterbalanced using a Latin
Square design to reduce the risk of order effects. Time taken to complete the
22
measures ranged from 30-60 minutes. Details of participants’ violent
forensic history, as well as details of violent incidents on the ward over the
past 12 months prior to the date of study completion, were obtained from
patients’ medical notes, then were counted and scored using the MOAS.
Patients were compensated £3 for their time. Once all data had been
obtained and analysed, participants were told they would be given a
summary sheet explaining the overall findings of the study if they indicated
that they wanted one on the consent form.
Data Analysis Strategy
Firstly correlational analyses were performed to explore relationships
between all variables.
Hypothesis 1
Regression analyses were conducted to test whether implicit and explicit
self-esteem, as well as their interaction were associated with each
aggression variable (see Table 4 with different types of regression according
to the properties of each outcome variable). Follow-up simple slopes tests
(Aiken & West, 1991) were conducted to explore the direction of a
significant explicit and implicit self-esteem interaction. Fragile self-esteem
was defined as a significant interaction between ESE and ISE, with simple
slopes tests conducted to explore whether high ESE and low ISE predicted
aggression. This allowed Hypothesis 1, that individuals with fragile self-
esteem (high explicit, low implicit) will have higher aggression ratings, to
be tested. An increase in aggression with a decrease in ISE but only when
ESE is high would indicate support for this hypothesis.
23
Hypothesis 2
Correlational and regression analyses suitable for each dependent variable
(see Table 4) were performed to test whether psychopathy predicted higher
levels of aggression.
Hypothesis 3
A logistic regression analysis was performed to test Hypothesis 3; that
psychopathy will predict instrumental (versus reactive) aggression in the
index offence.
Hypothesis 4 and 5
Multiple linear regression analyses were conducted to test whether implicit
and explicit self-esteem, as well as their interaction were associated with (a)
psychopathy and (b) narcissism (testing Hypotheses 4 and 5 respectively).
The direction of significant interactions was explored by conducting follow
up simple slopes analyses (with high explicit/low implicit representing
fragile self-esteem). An increase in narcissism/psychopathy with a decrease
in ISE but only when ESE is high would indicate support for these
hypotheses. Further identical regression analyses were then conducted to
look at the specific factors of narcissism (i.e. adaptive and maladaptive) and
psychopathy (i.e. Factor 1 and 2).
24
Table 4Table of regression analyses run for each aggression dependent variableDependent Variable Type of Variable Regression Model
Number of physically aggressive convictions
Count (no zero values)
Negative Binomial Poisson Regression
Severity of aggression in index offence
Ordinal categorical (low, medium or high)
Ordinal Regression
Involvement in a physically aggressive incident in the past 12 months
Binary (yes or no) Logistic Regression
Reactive or instrumental aggression involved in index offence
Binary (reactive or instrumental)
Logistic Regression
Results
Missing Data
Self-report measures were completed by all participants except for two NPIs
and one SRPS. Medical record information was not adequate to score
measures of reactive/instrumental aggression for three participants, and
PCL-R scores for Factor 1 and Factor 2 psychopathy were not available for
six individuals. One participant’s data was excluded from physically
aggressive convictions regression analyses as their score was greater than
3.29 standard deviations from the mean (Field, 2013), however all other
data for this individual was included in analysis.
Descriptive Statistics
The means and standard deviations for the key variables are presented in
Table 5. Internal consistency was also calculated for all independent
variables except the PCL-R because scores for individual items were not
available in the notes of some participants. All measures demonstrated good
25
internal consistency except for the IAT which was considered poor (George
& Mallery, 2003).
All variables were initially inspected for violations of the assumption of
normal distribution (Tabachnick & Fidell, 2007, see Appendix 7). As PCL-
R scores were not normally distributed and no data transformation could
produce an approximate normal distribution of these data, non-parametric
correlational analyses were performed. It was decided to split the number of
incidents into a binary variable (no incidents or 1 or more incidents in the
past 12 months) as 70% of the sample had not been physically aggressive on
the ward in the past 12 months. Ratings of reactive and instrumental
aggression involved in the index offence were categorized into two groups
(mostly instrumental and mostly reactive) in order to make the numbers of
participants in each group more even. Ratings of the severity of the index
offence were also categorized into three severity groups (low, medium and
high) because 46% of participants obtained the same severity score (see
Frequency Tables in Appendix 8).
A significant positive correlation was found between clinician-rated and
self-report total psychopathy (rs=.402, n=49, p<.01), however the strength
of this relationship was medium (Cohen, 1988). This finding suggests that
the SRPS and PCL-R may be measuring related but different constructs,
which is consistent with other findings comparing self-report and clinician-
rated psychopathy in incarcerated samples (Brinkley, Schmitt, Smith &
Newman, 2000). There was also reason to question the validity of self-
report psychopathy scores, as some participants stated that they were not
completing this measure truthfully. Therefore all subsequent analyses were
26
performed using PCL-R scores only, which is a well validated measure of
psychopathy (Butcher, 2009; Hare 2003).
Table 5 Means, standard deviations and alpha coefficients
Scale Mean SD Range Alpha % missingRSES 19.54 6.06 3-30 .89 0IAT .461 .383 -0.59-1.17 .54 0NPI 12.75 7.20 0-29 .87 2SRPS 56.27 10.55 39-79 .83 1PCL-R 22.2 7.84 4-36 - 0Convictions 7.14 6.73 1-28 - 0MOAS IO 15.42 5.66 0-26 - 0Incidents 0.74 2 0-12 - 0Instrumental / reactive
2.83 1.07 1-4 - 6
IAT Data
Mean accuracy rates for blocks 1-5 were 96.4%, 96.6%, 95%, 90.6% and
81.1% respectively. A one-way repeated measures ANOVA revealed a
significant difference in accuracy scores across blocks (Wilks’ Lambda=.48,
F(4,46)=12.6, p<.005, multivariate partial eta squared=.52). Greenwald and
Farnham (2000) recommended that participants who have error rates of 20%
or more should be excluded from analyses. However, given that 19
participants (38%) had overall mean error rates of over 20% it was decided
to retain all participants regardless of accuracy. It was difficult to compare
mean D scores between studies as the number of included blocks has varied
and a range of methods have been used to determine the IAT score (Gregg
& Sedikides 2010; Greenwald & Farnham, 2000; Greenwald McGhee &
Schwartz, 1998; Greenwald, Nosek & Banaji, 2003, see Appendix 5).
Nevertheless, the combination of lower internal consistency (r=.54) and
27
high errors rates suggest that the results in relation to the IAT should be
interpreted with caution.
Correlations Between Variables
As a first stage of this analysis, correlational analysis was performed to
explore the association between all variables. Table 6 presents inter-
correlations between all variables.
28
Table 6 Spearman’s rho correlations between variablesNote: SE= self-esteem, PCL-R= Psychopathy Checklist, F1= Factor 1, F2= Factor 2 1 2 3 4 5 6 7 8 9 10 11 12 13 M SD
Age (1) - .000 -.26 -.25 -.28 -.27 -.14 -.06 -.06 .17 -.10 -.05 .39** 39.34 8.97
Explicit Self-Esteem (2) .00 - .01 .46** .14 .41** .16 -.07 .08 -.06 -.15 .09 .15 19.54 6.06
Implicit Self-Esteem (3) -.26 .01 - .19 .06 .11 .11 -.02 -.04 .10 .04 -.19 -.09 .46 .38
Adaptive Narcissism (4) -.25 .46** .19 - .61** .89** .16 .02 .13 .04 .17 .05 -.07 5.58 3.41
Maladaptive Narcissism (5) -.28 .14 .06 .61** - .82** .39* .21 .35* .23 .30* .05 -.02 4.50 3.03
Narcissism (6) -.27 .41** .11 .89** .82** - .34* .11 .29* .12 .28 .09 -.08 12.75 7.20
Psychopathy F1 (7) -.14 .16 .11 .16 .39* .34* - .42** .78** .20 .17 .08 -.03 7.93 3.57
Psychopathy F2 (8) -.06 -.07 -.02 .02 .21 .11 .42** - .85** .44** .49** .13 .12 12.52 5.04
Psychopathy Total (9) -.06 .08 -.04 .13 .35* .29* .78** .85** - .36* .42** .12 .03 22.20 7.84
Aggressive Convictions
(10).17 -.06 .10 .04 .23 .12 .20 .44** .36* - .14 .17 .11 7.14 6.73
Number of incidents (11) -.10 -.15 .04 .17 .30* .28 .17 .49** .42** .14 - -.13 -.02 .74 2.0
Severity of Index Offence
(12)-.05 .09 -.19 .05 .05 .09 .08 .13 .12 .17 -.13 - -.18 15.42 5.66
Reactive/Instrumental
Aggression (13).39** .15 -.09 -.07 -.02 -.08 -.03 .12 .03 .11 -.02 -.18 - 2.83 1.07
*p≤.05 **p≤.01
29
Consistent with previous findings (Bosson et al., 2000; Hofmann et al.,
2005; Krizan & Suls, 2008), implicit (ISE) and explicit self-esteem (ESE)
were not significantly correlated.
Total narcissism was significantly, positively correlated with PCL-R Factor
1 and total scores but not Factor 2 scores. On closer inspection of the two
facets of narcissism, maladaptive narcissism correlated only with PCL-R
Factor 1 and total scores, whereas adaptive narcissism was uncorrelated
with all psychopathy indices. Adaptive narcissism was significantly,
positively correlated with ESE. There were some significant, positive
associations between narcissism, particularly maladaptive narcissism, and
aggression in terms of the number of physically aggressive incidents (see
Table 6).
Clinician-rated psychopathy was associated with a number of aggression
variables, which lends preliminary support to Hypothesis 2. Total
psychopathy was significantly positively correlated with number of
incidents and number of convictions but not the severity of index offence
(see Table 6). Closer inspection of the two psychopathy factors revealed that
Factor 2 but not Factor 1 psychopathy was significantly, positively
correlated with these aggression outcomes, with stronger relationships
compared to total psychopathy. Factor 1 was only weakly and non-
significantly associated with convictions and number of incidents.
Regression Analyses
In the next stages of the analysis, regression analyses were performed on the
dataset. All measures of self-esteem, narcissism and psychopathy were
30
centered around their means prior to being entered into regression models
(Cohen, Cohen, West & Aiken, 2003). This was to ease interpretation of the
data, as well as making the scales of the dependent and independent
variables comparable (Aiken & West, 1991).1
Hypothesis 1: Self-Esteem Fragility and Aggression
Hypothesis 1; that self-esteem fragility will predict higher aggression
ratings, was tested using various regression analyses suitable for the nature
of the four aggression dependent variables (i.e. poisson, logistic and ordinal
regression), entering ESE, ISE and the interaction between these two
variables as continuous predictors.
Self-esteem fragility and aggressive convictions
To explore the hypothesis that self-esteem fragility will predict aggression
in relation to physically aggressive convictions, a zero-truncated negative
binominal regression was used with number of physically aggressive
convictions as the dependent variable and with ESE, ISE and their
interaction as continuous predictors (see Table 7). The model as a whole
was not statistically significant (LR χ2(3, N = 50) = 2.08, p = .555),
indicating that discrepant self-esteem did not predict a higher log count of
physically aggressive convictions. None of the individual variables
(including the interaction) made a significant contribution to the model,
therefore there was no support for the hypothesis that fragile self-esteem
would predict aggressive convictions. The alpha dispersion parameter
1 Regression analyses were also conducted for physically aggressive convictions excluding robbery (Appendix 10) as definitions of aggressive convictions vary across the clinical and legal literature.
31
confirms that a negative binomial model provides a better estimation of the
data than a zero truncated poisson model.
Table 7Zero truncated negative binomial regression predicting physically aggression convictions including robbery using explicit and implicit self-esteem
Coefficient Standard Error
95% confidence intervals P-value Pseudo R2
Likelihood ratio χ2
alpha = 0 (p)
Explicit SE -.18 .17 -.52 to .16 .29
.01 138.81(p<.001)
Implicit SE .17 .16 -.14 to .49 .29Explicit*Implicit SE
.027 .24-.44 to .49
.91
Alpha 1.04 .42 .47 to 2.27 -Note: SE=self-esteem
Self-esteem fragility and Incidents on the Ward
To test Hypothesis 1 in relation to institutional aggression, a logistic
regression was performed with group membership (no incidents in the past
12 months or 1 or more incidents in the past 12 months) as the binary
dependent variable, with ESE, ISE and the interaction between these two
variables as continuous predictors (see Table 8). The full model was not
statistically significant, (χ2(3, N = 50) = 2.513, p = .473), indicating it was
unable to distinguish between those who had and had not been involved in a
physically aggressive incident in the past 12 months. Thus there was no
support for the hypothesis fragile self-esteem predicts institutional
aggression.
32
Table 8Logistic regression predicting institutional aggression using explicit and implicit self-esteem
Beta Exp(B) Wald 95% confidence
intervals P-value Nagelkerke R Square
Explicit SE -.47 .63 2.02 .33 to 1.19 .16.07Implicit SE .24 1.27 .50 .66 to 2.44 .48
Explicit*Implicit SE -.03 .97 .01 .45 to 2.12 .94
Note: SE=self-esteem
Self-esteem fragility and Severity of Index Offence
To explore the Hypothesis 1 in relation to the severity of aggression in the
index offence, an ordinal regression analysis was performed to determine
whether the odds of committing a more aggressive index offence differed
significantly according to levels of ESE, ISE and discrepant (the
explicit*implicit interaction) self-esteem (see Table 9). Index offence
severity (low, medium, high) was the ordinal dependent variable. The test of
parallel lines was not significant (χ2(3) = 1.258, p = .739) indicating that
each independent variable had an identical effect at each cumulative split of
the ordinal dependent variable, thus the assumption of proportional odds had
been met. The full model containing all predictor variables was not a
significantly better fit than the model without the predictor variable (χ2(3) =
4.289, p = .232). The model as a whole explained 9% of the variance in
group membership. None of the individual variables nor the interaction
effect were significant. Thus there was no support for the hypothesis that
fragile self-esteem predicts higher aggression in terms of the severity of the
index offence.
33
Table 9Ordinal regression model severity of index offence using explicit and implicit self-esteem
Estimate Wald 95% confidence intervals
P-value
Nagelkerke Pseudo R Square
Threshold MOAS Group 0
-1.45 15.81 -2.17 to -.74 .00
.09MOAS Group 1
.73 5.51 .12 to 1.34 .02
Location Explicit SE .23 .65 -.32 to .77 .42Implicit SE -.21 .55 -.75 to .34 .46Explicit* Implicit SE
.63 2.75 -.12 to 1.38 .10
Note: SE=self-esteem
Because ESE was significantly correlated with narcissism, both of these
variables were entered into a regression model, alongside ISE and the
ESE*ISE interaction for each of the four aggression outcome variables as a
post-hoc analysis (see Appendix 9). Only the model predicting the
likelihood of aggressive incidents was significant (χ2(48, 4)=.9.826, p<.05).
Individually, higher narcissism (β =1.050, p<.05) and lower levels of ESE (β
=-.995, p<.05) made significant unique contributions to the model,
indicating that these two variables are associated with institutional
aggression. Comparison of the model of aggressive incidents including and
excluding narcissism indicates that the former explains 13.6% (R2 was .136
higher) more variance in convictions. The same regression analyses were
performed with adaptive and maladaptive narcissism as outcomes and no
significant results were found (see Appendix 9).
Hypothesis 2: Psychopathy & Aggression
Psychopathy was entered into regression models, alongside ESE, ISE and
the ESE*ISE interaction for each aggression outcome. The models
predicting the likelihood of aggressive incidents (χ2(50, 4)=.15.558) and
counts of aggressive convictions LR χ2(4, N = 50)=11.96) were significant
34
(p<.05), with psychopathy the only variable to make significant unique
contributions (β=1.596, p<.01 and β =.571, p=.001 respectively). The
models containing psychopathy as predictors explained additional 21.8% of
the variance (R2 was .218 higher) in incidents and 5% in convictions (see
Appendix 10), which supports Hypothesis 2, that psychopathy will predict
higher levels of aggression.
Hypothesis 3: Psychopathy and Type of Aggression
To test Hypothesis 3; that psychopathy will predict more instrumental
aggression in the index offence a logistic regression was performed with
group membership (instrumental or reactive) as the binary dependent
variable and PCL-R scores as the continuous predictor variable.
The full model was not statistically significant, χ2(1, N = 47) = 1.114, p
= .291) thus there was no support for the hypothesis that psychopathy
predicts more instrumental aggression in the index offence (see Table 11).
Table 11Three logistic regressions predicting type of violence in index offence using psychopathy (first row), psychopathy factor 1 (second row) and factor 2 (third row)Outcome variable of regression model
Beta Exp(B) Wald 95% CI P-value Nagelkerke R Square
Total Psychopathy .33 1.387 1.087 0.75-
2.56 .30 .03
Psychopathy Factor 1
.23 1.25 .51 .67-2.34 .48 .02
PsychopathyFactor 2
.52 1.68 2.11 .84-3.36 .15 .07
Note: SE=self-esteem
35
Further identical regression analyses were performed to explore whether
individual factors of psychopathy (factors 1 and 2) were related to
instrumental and reactive aggression. None of these analyses were
significant (see the lower two rows of Table 11).
36
Table 12Linear regression predicting total, factor 1 and factor 2 psychopathy using explicit and implicit self-esteem
Total Psychopathy Factor 1 Psychopathy Factor 2 Psychopathy
β (B) 95% CI P Value Adjusted R2
β (B) 95% CI P Value Adjusted R2
β (B) 95% CI P Value Adjusted R2
Explicit SE
.05 (0.37) -1.89-2.63 .74
0.25
.13 (0.46) -0.62-1.54 .40
0.07
-.08 (-0.42) -1.98-1.15 .59
0.02Implicit SE
-.00 (-0.03) -2.28-2.22 .98 .11 (0.38) -0.70-1.45 .48 -.01 (-0.04) -1.59-1.52 .96
Explicit*Implicit SE
.30 (2.83) 0.05-5.61 .05 .36 (1.57) 0.24-2.89 .02 .26 (1.63) -.30-3.55 .10
Note: SE=self-esteem
37
Hypothesis 4: Self-esteem Fragility and Psychopathy
Identical regression analyses were conducted with PCL-R total, Factor 1 and
Factor 2 score as the continuous outcome variables to test Hypothesis 4; that
fragile self-esteem will predict higher levels of psychopathy. The models
did not reach statistical significance (R Square=.025, F(3,46)=1.421,
p=.249). ESE and ISE alone were not significantly associated with total
psychopathy however the interaction between them was (β=2.826, p=.046),
suggesting that the effect of ESE on total psychopathy depended on ISE
score (see Table 12).
Follow up simple slopes tests (Aiken & West, 1991) were performed to
determine the direction of the ISE/ESE interaction (with high ESE and low
ISE representing fragile self-esteem) in relation to total psychopathy.
Simple slopes for the association between ESE and total psychopathy were
tested for low (-1 SD below the mean) and high (+1 SD above the mean)
levels of ISE. This model is the most common method used in studies
exploring the fragility of self-esteem (Jordan, Spencer, & Zanna, 2005;
Jordan, Spencer, Zanna, Hoshino-Browne, & Correll, 2003; Kernis et al.,
2005).
38
Figure 1. Psychopathy as a Function of Explicit & Implicit Self-Esteem
The interaction of ISE and ESE predicting overall psychopathy is presented
in Figure 1. Among patients low in ESE (-1 SD; left side of Figure 1), there
was a slight but non-significant negative relationship between ISE and
psychopathy (B=-.405, t=-1.496, p=.141). Among patients high in ESE (+1
SD; right side of Figure 1), there was a slight but non-significant positive
relationship between ISE and psychopathy (B=.528, t=1.676, p=.10). These
findings do not support Hypothesis 4, instead they suggest that patients with
congruent self-esteem have a tendency to be more psychopathic compared
to people with other combinations of ESE/ISE; however this relationship
was not statistically significant. Nevertheless this trend highlights the
possibility of a difference in self-esteem profiles of narcissistic and
psychopathic individuals. A moderation analysis was performed using a
PROCESS macro code for SPSS (Version 2.15, Hayes 2013) in order to
explore the moderating role of ISE on the relationship between ESE and
psychopathy. The results revealed the proportion of variance in psychopathy
39
uniquely attributable to the moderation of ESE’s effect by ISE was 8.3%
(R2-change=.083, F(1,46)=4.19, p=.046).
A further linear regression analysis was run to explore the self-esteem
profile of psychopathy in more depth, in relation to F1 and F2 of this
construct (PCL-R). The overall models did not reach statistical significance.
ESE and ISE alone were not significant individual predictors however the
interaction between them was (β=1.57, p=.02) for Factor 1 psychopathy
only, suggesting that the effect of ESE on psychopathy factor 1 depended on
the level of ISE (see Table 12).
Figure 2. Psychopathy Factor 1 as a Function of Explicit & Implicit Self-Esteem
Follow up simple slopes tests were performed to determine the direction of
the ISE/ESE interaction (with high ESE and low ISE representing fragile
self-esteem) in relation to Factor 1 psychopathy. The interaction of ISE and
ESE predicting overall Factor 1 psychopathy is presented in Figure 2. A
similar pattern was found with patients low in ESE (-1 SD; left side of
Figure 2); there was a slight but negative relationship between ISE and
40
psychopathy Factor 1 although this was not significant (B=-.165, t=-1.351,
p=.184). Among patients high in ESE (+1 SD; right side of Figure 2), there
was a significant positive relationship between ISE and Factor 1
psychopathy (B =.361, t=2.289, p=.0274). Therefore individuals with high
congruent self-esteem appear to have higher levels of Factor 1 psychopathy.
41
Table 13Linear regression predicting total, adaptive and maladaptive narcissism using explicit and implicit self-esteem
Total Narcissism Adaptive Narcissism Maladaptive Narcissism
β (B) 95% CI P Value Adjusted R2
β (B) 95% CI P Value Adjusted R2
β (B) 95% CI P Value Adjusted R2
Explicit SE
.41 (2.91) 0.91-4.91 .01
.14
.47 (1.59) 0.70-2.49 .00
.22
.10 (0.31) -0.61-1.22 .50
-.02Implicit SE
.06 (0.42) -1.57-2.41 .67 .13 (0.45) -0.45-1.34 .25 -.02 (-0.06) -0.97-0.86 .91
Explicit*Implicit SE
-.09 (-0.77) -3.22-1.69 .53 -.10 (-0.40) -1.50-0.71 .47 -.16 (-0.60) -1.72-0.54 .30
Note: SE=self-esteem
42
Hypothesis 5: Self-esteem fragility and narcissism
Hypothesis 5, that fragile self-esteem will predict higher levels of
narcissism, was tested using a multiple linear regression with total
narcissism as the continuous outcome variable. ISE, ESE, and their
interaction term (implicit * explicit) were the continuous predictor variables.
The model as a whole significantly predicted variance in narcissism (F(3,
44)=3.465, p<.05), with 14% of the variance in narcissism scores was
explained by the model. Only ESE made a statistically significant unique
contribution (β=.405, p<.01), controlling for the other included variables
(see Table 8), with ESE uniquely explaining 16.4% of the variance (squared
partial correlation=.164). This suggests that, as ESE increased, so did total
narcissism.
Further regression analyses with ISE, ESE, and their interaction term
(implicit * explicit) revealed that ESE made a statistically significant unique
contribution to a model of adaptive narcissism (F(3, 44)=5.7, p<.01),
explaining 22% of the variance in adaptive narcissism (squared partial
correlation=.23). Only ESE made a significant unique contribution (β=.47,
p<.01) explaining 23% of the variance. This finding was not replicated with
maladaptive narcissism (see Table 13). These findings do not support
Hypothesis 5, that fragile self-esteem predicts higher levels of narcissism,
instead it suggests that high ESE predicts variance in total and adaptive
narcissism, which is consistent with some of the previous findings in the
literature (Bosson et al., 2008; Bushman & Baumeister, 1998).
Discussion
43
The relationship between self-esteem and aggression is unclear, and has
mainly been explored so far in non-clinical populations only. This study is
one of the first to explore self-esteem fragility and direct, physical
aggression in a population of mentally disordered offenders who are at high
risk of harming others; an under-researched and hard to treat group. It
represents the first step in an important journey and makes a vital
contribution to the evidence base by measuring aggression in a variety of
ways and by highlighting methodological issues in measuring self-esteem in
such a unique population. The primary aim of the present study was to
explore the relationship between self-esteem fragility and aggression.
Further aims were to understand the self-esteem profile of narcissism and
psychopathy.
Self-Esteem and Aggression
Overall, findings did not support the hypothesis that fragile self-esteem will
predict higher levels of aggression. However it was found that low ESE and
high narcissism may play a role in predicting the risk of aggressive
behaviour in an institutional environment. This finding is not consistent with
those from another study exploring aggression in a psychiatric inpatient
setting (Goldberg et al. 2007), which found that high ESE and high
narcissism (using the same measures as those used in the present study)
were associated with aggression. These findings also do not support the
Threatened Egotism Hypothesis (Baumeister, et al., 1996; Baumeister, et al.,
2000), but lend some support to “mask models” of narcissism, self-esteem
and aggression (Kernberg, 1975; Kohut, 1966). In a high secure context,
narcissism may serve the function of regulating self-esteem, however when
44
exposed to an ego threat and ESE is damaged, individuals may confront
others in a particular way (for example with machismo) as a way of
protecting their self-esteem, and this pattern of relating may be associated
with aggression (Walker, 2005; Walker & Gudjonsson, 2006). Indeed,
narcissism has been associated with emotion dysregulation and externalising
behaviours such as aggression in other research (Lau & Marsee, 2013;
Baumeister et al., 2000; Washburn et al., 2004). This way of relating to
others could be especially functional in a high secure environment, when an
individual may need to come across in a particular way in order to protect
themselves from threats and humiliation (Nestor, 2002; Svindseth et al.,
2006). In the absence of functional ways of managing distress individuals
may express it in uncontained ways e.g. through violence. It is unclear why
this pattern was not found with maladaptive narcissism however it is
possible that this analysis was underpowered.
The majority of incidents reported in the present study were for verbal
aggression, and few incidents were reported for physical aggression in the
past 12 months. It has been highlighted that aggression in inpatient settings
is under-reported in clinical records (Brin et al., 2000), therefore the data
obtained from medical notes may not have captured the true frequency of
incidents.
The low accuracy rates and poor internal consistency in this study for the
IAT suggest that it may have been measuring some other construct
(Tafardoi & Ho, 2006). The presence of psychiatric symptoms may have
influenced performance on the IAT, for example by affecting
comprehension and concentration which would affect reaction times (Gratz,
45
2016). Further, side effects of psychiatric medication could have slowed
reaction times (Plesnicar, Zalar, Tomori & Krajnc, 2003), and
neuropsychological difficulties common in people treated in high secure
settings (Lowings, 2010; Williams et al., 2010), as well as a lack of
familiarity with computers, may have contributed to participants’ high error
rates and long latencies on the IAT in the present study. This task was
originally validated with student samples who are likely to be used to using
computers in contrast to highly institutionalised individuals. Thus the
influence of ISE on aggression cannot be discounted. Measures of ISE that
do not rely on reaction time may be more appropriate measures to use in a
high secure psychiatric population to give a better idea of the relationship
between ISE, a component of self-esteem fragility, and aggression.
It is difficult to compare these findings with those from other studies due to
the different measures of ISE used, various methods used to score the IAT
and various contexts in which ISE measures are administered (Jordon et al.,
2003; Gregg & Sedikides, 2010; Vater et al., 2013). It has been
demonstrated that the testing environment and words used can affect the
results of implicit measures (see Blair et al., 2002 for a review; (Karpinski,
2004). It is possible that variations in how ISE measures are developed,
administered and scored are partly responsible for the inconsistent findings
in the evidence base. Nevertheless, implicit measures do need to be adapted
for the client group with which they will be used.
Psychopathy and Aggression
In forensic psychiatric populations, there may be factors other than self-
esteem that predispose someone to act aggressively, which might include
46
personality factors such as psychopathy. Consistent with previous findings
(Blackburn & Coid, 1998; Herve, Hayes & Hare, 2003; Toch, et al.,
1989; Snowden et al., 2004), the hypothesis that psychopathy will predict
higher levels of aggression, was supported in terms of previous convictions
and incidents but not severity of index offence. A reason for this may be that
many participants obtained the same severity rating for their index offence
(46%). Although the scores on this measure were categorized into groups
rather than treated as a continuous outcome variable, the way in which the
groups were split may have meant there was little variation between groups.
However others have also found that psychopathy is associated with
aggression and crime, but not with the type of seriousness of an offence
(Fougere, Potter & Boutilier, 2009). Therefore psychopathy may be
associated with a higher frequency of aggressive incidents and offences, but
not the severity of aggression involved.
Psychopathy predicted more variance in aggression than other factors such
as self-esteem and narcissism. A closer examination of the individual factors
of psychopathy revealed a stronger relationship between Factor 2, compared
to total psychopathy, and previous convictions and institutional aggression.
This could be because the particular traits of Factor 2 psychopathy; a history
of anti-social behaviour and impulsivity, have been linked to a higher risk of
aggression in the literature (see Gvion & Apter, 2011 for a review; Bonta,
Law & Hanson, 1998).
Psychopathy is considered a stable personality construct that is resistant to
change, even following treatment, although this is contentious (Harpur &
Hare, 1994; Polaschek, 2014). This could by why this particular construct
47
had a stronger relationship with aggression in this study, compared to other
constructs which were measured which could have been modified by
treatment.
Evidence suggests that psychopathy is more related to instrumental
aggression rather than reactive aggression (Cornell et al., 1996; Woodworth
& Porter, 2002), however this hypothesis was not supported in the current
study. Further analysis of the two factors of psychopathy revealed no
significant relationship between either and the type of aggression in the
index offence.
Self-Esteem and Psychopathy
This is the first piece of research that has directly studied the relationship
between fragile self-esteem and psychopathy. Research into the link
between shame and violence indicates that psychopathic individuals use
dominance and aggression to protect their vulnerable core self from feelings
of humiliation (Morrison & Gilbert, 2001; Blackburn, 1975), which may
represent fragile self-esteem. However the hypothesis that self-esteem
fragility will predict higher levels of psychopathy was not supported.
Unexpectedly, congruent (high explicit and high implicit) self-esteem was
associated with Factor 1 psychopathy only which is characterized by high
levels of grandiosity and low levels of empathy. This finding is the first of
its kind and surprising given the current understandings of self-esteem in
psychopathic individuals (Kernberg, 1975; Cale & Lilienfeld, 2006).
Combined with the finding that ESE was positively associated with
narcissism, particularly adaptive, it raises the possibility that different
personality traits may have different self-esteem profiles. However future
48
research using a more appropriate measure of ISE would be of value to
explore this further in psychiatric samples. There are a number of ways of
conceptualising fragile self-esteem, of which this study explored only one.
Further research is needed to explore whether other types of fragile self-
esteem, for example self-esteem stability, to enhance our understanding of
the self-esteem profile of psychopathy.
Some have argued that self-report measures should not be used with
offenders to assess psychopathy (Cleckley, 1988; Copestake, Gray, &
Snowden, 2011; Ray et al., 2013), whereas others have found that
psychopathic offenders have good insight into these traits and report them in
line with clinician ratings when there are no perceived adverse
consequences (Miller, Jones & Lynam, 2011). Although the SRPS has been
reported as measuring a similar construct to the PCL-R by Brinkley et al.
(2001), they reported a similar small-moderate correlation to the one in the
present study, and one would expect to see a stronger relationship in
measures of a similar construct. In the present study the strength of the
correlation between self-report and clinician-rated psychopathy indicated
that they may have been measuring different constructs. Therefore future
research in high secure psychiatric settings should use a well-validated
measure to address hypotheses concerning psychopathy.
Self-Esteem and Narcissism
The hypothesis that fragile self-esteem will predict narcissism, was not
supported. Instead ESE alone predicted narcissism. This finding is
consistent with those of others (Brown & Zeigler-Hill, 2004; Sedikides et
al., 2004), and they extend them by showing that this relationship may also
49
be present in offenders with psychiatric diagnoses and a history of high
levels of physical violence, however further research with larger samples are
required to test this further. The findings of the present study are consistent
with those of Lima (2007) and Stoessel (2008), who also did not find a
fragile self-esteem profile in narcissistic individuals. However the
questionable validity of the IAT in measuring ISE means that one cannot
claim that ISE, and therefore fragile SE, is not associated with narcissism
and more research is required.
It is important to note that narcissistic traits rather than symptoms of
Narcissistic PD were measured in this study. Some have found that when
narcissism is measured using the NPI, it has a positive relationship with
ESE (Pincus et al., 2009; Svindseth et al., 2008), however when is it
assessed with symptom-based measures which supposedly capture
pathological narcissism (see Miller & Campbell, 2008; Vater et al., 2013), it
has a negative relationship with ESE in individuals with psychiatric
diagnoses (Vater et al., 2013). Although pathological, symptom-based
narcissism was not measured in the present study, maladaptive narcissism
was considered individually and it was not significantly associated with ISE
or ESE. However adaptive elements of narcissism were positively
associated with ESE. Thus research into the relationship between self-
esteem and narcissism may benefit from using measures that capture the full
complexity of these constructs.
Narcissism and Aggression
Measuring narcissism in more depth raised the possibility that
onlymaladaptive narcissism is positively associated with institutional
50
aggression, which is consistent with findings of others (Svindseth al., 2008;
Lau & Marsee, 2013; Edwards & Bond, 2012). Measures assessing different
sub-types of narcissism may be useful when assessing risk for violence in
clinical and forensic settings. The use of general measures of narcissism
may explain why there are mixed findings regarding the relationship
between narcissism and aggression.
Treatment Implications
These findings suggest that a more in depth assessment of self-esteem may
be beneficial in forensic psychiatric settings, which often focus only on
global self-reports (Carr & Browne, 2015). It is possible that, for some
people in institutional settings, narcissism could be functioning as a way of
regulating self-esteem which may not always be successful. Treatment
approaches that support the regulation of self-esteem and emotions, which
may be subject to fluctuation either in response to ego threats or due to
being detained in a high secure environment, may be beneficial in
developing more adaptive coping mechanisms (Linehan, 1993; Young et al.,
2006). Treatment recommendations cannot be made based on these findings
alone, especially as this study was cross-sectional in nature, and compared
present ratings of personality factors with ratings of previous aggressive
behaviour. However these findings, along with others, suggest that it may be
useful to consider psychopathy and narcissism alongside self-esteem when
assessing risk. Although it may be helpful to target offenders’ self-esteem in
treatment to enhance wellbeing, targeting other factors as well may reduce
the risk of aggression. Although psychopathy has been considered
“untreatable” (Hare et al., 2000; Hobson, Shine, & Roberts, 2000; Rice,
51
Harris, & Cormier, 1992; Seto & Barbaree, 1999), there is some evidence to
suggest that treatments for psychopathy can reduce recidivism (Skeem et al.,
2002). It has recently been demonstrated that treatments which focus on
core beliefs regarding the self, or “schemas”, can successfully reduce
institutional aggression and recidivism in forensic patients with Anti-Social
PD, Narcissistic PD and high levels of psychopathic traits (Bernstein et al.,
2012).
Evidence suggests that secondary psychopaths have lower self-esteem,
higher levels of emotional disturbance and are more aggressive compared to
primary psychopaths who are more confident and engage with others using
a dominant interpersonal style (Blackburn, 1998; Heilbrun & Heilbrun,
1985). Therefore the self-esteem profiles present in different types of
psychopathy may underlie the risk of aggression and may require different
approaches to treatment. Due to assumptions of the available mediation and
moderation statistical tools (Hayes, 2013), this idea could not be directly
tested in the present study due to the nature of the data collected. It is hoped
that, with the developments in mediation and moderation statistical tools,
this could be tested in the future. Clinicians may find it informative to assess
various facets of self-esteem, psychopathy and narcissism as they may
interact in a number of ways in relation to aggression and require different
treatment approaches.
In order to understand the aspects of self which contribute to psychological
distress as well as risk of aggression, more valid methods of measuring
implicit feelings of self-worth are required.
52
Limitations
These findings from a group of individuals who are at high risk of harming
others offer stronger ecological validity than the majority of studies with
non-clinical, usually student samples. Nevertheless these results are only
representative of a sub-population of one high secure hospital, where the
environment is designed to reduce aggressive behaviour. It is therefore not
typical of settings where violent behaviour may occur. Nevertheless, the
results will hopefully encourage further research to be conducted in other
clinical and forensic settings.
It is highly likely that clinicians only referred patients who were stable and
had not been involved in many recent incidents for the study, raising the
possibility of a selection bias which limits the generalizability of these
findings. It is also possible that aggression levels changed according to
patients’ journey throughout the hospital system. For example, research has
indicated that the majority of inpatient aggression occurs during the first few
weeks of hospital admission (Goldberg, 2007) and clinicians may not have
referred patients for research until their risk had been fully assessed and
treatment plan agreed. Therefore there is a possibility that these findings are
only representative of a sub-set of stable, settled patients at a particular stage
on their journey towards recovery. It is also possible that, as with the self-
report psychopathy measure, the self-report narcissism scores were
intentionally inflated by this group as a defence mechanism, for example
needing to come across in a particular way in order to communicate
wellness and progress.
53
Another limitation is that offenders were asked to rate how they currently
felt about themselves rather than how they felt at the time of the index
offence. Future research could ask participants to rate how they felt about
themselves at the time of the index offence, although their responses may be
influenced by memory and treatment effects.
It is possible that treatment received whilst in hospital enhanced both ISE
and ESE levels in this sample. Indeed, enhancing self-esteem is a treatment
target of many therapies used in secure psychiatric hospitals and with
offenders such as Schema Therapy and Dialectical Behaviour Therapy
(Young et al., 2006; Linehan, 1993), however it is unclear whether they
raise ESE, ISE, or both. The potential confounding effects of psychological
treatment should be considered when interpreting the findings from these
populations.
Conclusions
This study found that self-esteem fragility was not associated with
aggression in a sample of forensic psychiatric patients. It may be beneficial
to consider ESE alongside narcissism when predicting the risk of
institutional aggression as narcissism may be functioning as a way of
regulating ESE in a high secure environment. The results from this study
suggest that narcissism and psychopathy may have different self-esteem
profiles. Previous findings that psychopathy is associated with increased
levels of aggressive behaviour were supported, both in terms of previous
violence and institutional aggression, but not with severity of aggression in
the index offence. Considering a number of important, complex personality
factors associated with aggressive behaviour may aid formulations of the
54
causes of aggression in high secure settings. An effective formulation would
require an assessment of the various facets of self-esteem, narcissism and
psychopathy. More valid measures of ISE appropriate for us in the present
population are required.
55
References
Aiken, L. S., & West, S. G. (1991) Multiple Regression: Testing and
Interpreting Interactions. Newbury Park, CA: Sage.
American Psychiatric Association, (2013). Diagnostic and Statistical
Manual of Mental Disorders (5th ed). Arlington, VA: American
Psychiatric Publishing.
Barry, C.T., Frick, P.J., Adler, K.K., & Grafeman, S.J. (2007). The
Predictive Utility of Narcissism among Children and Adolescents:
Evidence for a Distinction between Adaptive and Maladaptive
Narcissism, Journal of Child & Family Studies, 16, 508–521.
Bateman, A.W., & Fonagy, P. (2004). Mentalization-based treatment of
BPD, Journal of Personality Disorders, 18, 36-51
Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001).
Bad is stronger than good. Review of General Psychology, 5, 323–
370.
Baumeister, R. F., Bushman, B. J., & Campbell, W. (2000). Self-esteem,
narcissism, and aggression: Does violence result from low self-
esteem or from threatened egotism? Current Directions in
Psychological Science 9(1), 26-29.
Baumeister, R. F., Smart, L. & Boden, J.M. (1996). Relation of threatened
egotism to violence and aggression: The dark side of high self-
esteem, Psychological Review 103(1), 5-33.
Bentall, R.P., Kinderman, P., & Kaney, S. (1994). The self, attributional
processes and abnormal beliefs: Towards a model of persecutory
delusions, Behaviour, Research & Therapy, 32, 331–341.
56
Bernstein, D.P., Nijman, H., Karos, K., Keulen-de Vos, M., de Vogel, V., &
Lucker, T. (2012). Schema Therapy for forensic patients with
personality disorders: Design and preliminary findings of
multicenter randomized clinical trial in the Netherlands.
International Journal of Forensic Mental Health, 11, 312-324
Blackburn, R. (1975). An empirical classification of psychopathic
personality, British Journal of Psychiatry 127, 456–60.
Blackburn, R., & Coid, J.W. (1998). Psychopathy and the dimensions of
personality disorder in violent offenders, Personality and Individual
Differences, 25(1), 129-145.
Blair, I. (2002). The malleability of automatic stereotypes and prejudice.
Personality and Social Psychology Review, 6, 242-260.
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.
Borton, J.L.S., Crimmins, A.E., Ashby, R.S., & Ruddiman, J.F. (2012).
How do individuals with fragile self-esteem cope with intrusive
thoughts following ego threat? Self & Identity, 11(1), 16-35.
Bosson, J.K., Lakey., C.E., Campbell, W.K., Zeigler-Hill, V., Jordan, C.H.,
& Kernis, M.H. (2008). Untangling the Links between narcissism
and self-esteem: A theoretical and empirical review, Social and
Personality Psychology Compass 2/3, 1415–1439.
Bosson, J. K., Swann, W. B., & Pennebaker, J. W. (2000). Stalking the
perfect measure of implicit self-esteem: The blind men and the
57
elephant revisited? Journal of Personality and Social Psychology,
79, 631-643.
Bradley, M.M. & Lang, P.J. (2010). Affective Norms for English Words
(ANEW): Instruction Manual and Affective Ratings. Technical
Report C-2. University of Florida, Gainesville, FL.
Brinkley, C. A., Schmitt, W.A., Smith, S. S., & Newman, J. P. (2001).
Construct validation of a self report psychopathy scale: Does
Levenson’s SRPS measure the same construct as Hare’s PCL–R?
Personality and Individual Differences, 31, 1021-1038.
Brown, R.P., & Zeigler-Hill, V. (2004). Narcissism and the non-equivalence
of self-esteem measures: A matter of dominance? Journal of
Research in Personality, 38, 585–592.
Bushman, B. J. & R. F. Baumeister (1998). Threatened egotism, narcissism,
self-esteem, and direct and displaced aggression: Does self-love or
self-hate lead to violence?, Journal of Personality and Social
Psychology 75(1), 219-229.
Butcher, J.N. (2009). Oxford Handbook of Personality Assessment, NY:
Oxford University Press.
Cafri, G., van den Berg, P., & Brannick, M.T. (2010). What have the
difference scores not been telling us? A critique of the use of self-
ideal discrepancy in the assessment of body image and evaluation of
an alternative data-analytic framework, Assessment, 17(3), 361-376.
Cale, E.M., & Lilienfeld, S.O. (2006). Psychopathy factors and risk for
aggressive behaviour: A test of the ‘‘threatened egotism’’
hypothesis. Law and Human Behaviour, 30, 51–74.
58
Campbell, J.D., Assand, S., & Di Paula, A. (2000). Structural features of the
self-concept and adjustment, In A. Tesser, R.B. Felson, & J.M. Suls,
(Eds). Psychological Perspectives on Self and Identity (pp 115-138).
Washington, DC, US: American Psychological Association.
Campbell, K.W., & Foster, J.D. (2007). The narcissistic self: Background,
an extended agency model, and ongoing controversies. In C.
Sedikides & S.J. Spencer (Eds) Frontiers in Social Psychology: The
Self (pp 115-138). UK: Psychology Press.
Carr, M., & Browne, K.D. (2015). Measurement and conceptualisation of
self-esteem with forensic clientele, Assessment & Development
Matters, 7(3), 20-22.
Cleckley, H. (1941) The So-called Psychopathic Personality, With Special
Emphasis on his Status in the Selective Service, Journal of the
Medical Association of Georgia, 30, 466–472.
Cleckley, H.M. (1988). The Mask of Sanity: An Attempt to Clarify Some
Issues About the So Called Psychopathic Personality, William a
Dolan; 5th edition.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences.
Hillsdale, NJ: Erlbaum.
Cohen, J., Cohen, P. C., West, S. G., & Aiken, L. S. (2003). (3rd Ed.)
Applied Multiple Regression/Correlation Analysis for the Behavioral
Sciences. Mahwah, NJ.: Lawrence Erlbaum.
Coid, J.W., & Roberts, A. (2007). Psychopathy and offending behaviour:
Findings from the National Survey of Prisoners in England & Wales,
Journal of Forensic Psychiatry and Psychology, 18, 23-43.
59
Copestake, S., Gray, N. S. and Snowden, R. J. (2011). A comparison of a
self-report measure of psychopathy with the psychopathy checklist-
revised in a UK sample of offenders, Journal of Forensic Psychiatry
& Psychology, 22(2), 169-182.
Cornell, D. G., Warren, J., Hawk, G., Stafford, E., Oram, G., & Pine, D.
(1996). Psychopathy of instrumental and reactive violent offenders.
Journal of Consulting and Clinical Psychology, 64, 783-790.
Creemers, D.H.M., Schlte, R.H.J., Engels, R.C.M.E., Prinstein, M.J &
Wiers, R.W. (2013). Damaged self-esteem is associated with
internalising problems, Frontiers in Psychology, 4, Article 152.
Department of Health. (2003). Personality disorder: No Longer a Diagnosis
of Exclusion. London: National Institute for Mental Health.
DeWall, C.N. & Anderson, C.A. (2011). The general aggression model. In
P.R Shaver & M. Mikulincer (Eds.), Human Aggression & Violence:
Causes, Manifestations & Consequences (pp. 15-33). Washington,
DC, USA: American Psychological Association.
Donnellan, M.B., Trzesniewski, K.H., Robins, R.W., Moffitt, T.E., & Caspi,
A. (2005). Low Self-Esteem Is Related to Aggression, Antisocial
Behavior, and Delinquency, Psychological Science, 16(4), 328-335.
Edwards, J.R. (2001). Ten difference score myths, Organizational Research
Methods, 4(3), 265-287.
Edwards, J. R. (2002). Alternatives to difference scores: Polynomial
regression and response surface methodology. F. Drasgow & N.W
Schmitt (Eds.), Advances in Measurement and Data Analysis
(pp.350-400). San Francisco: Jossey-Bass.
60
Edwards, R. & A. J. Bond (2012). Narcissism, self-concept clarity and
aggressive cognitive bias amongst mentally disordered offenders,
Journal of Forensic Psychiatry & Psychology 23(5-6), 620-634.
Falkenbach, D.M., Howe, J.R., & Falki, M. (2013). Using self-esteem to
disaggregate psychopathy, narcissism and aggression, Personality &
Individual Differences, 54, 815-820.
Farrington, D.P. 1998. Predictors, causes and correlates of male youth
violence. In Youth Violence, Crime and Justice, Vol. 24, M. Tonry
and M.H. Moore (Eds). Chicago, IL: University of Chicago Press,
pp. 421–475.
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A
flexible statistical power analysis program for the social, behavioral,
and biomedical sciences. Behavior Research Methods, 39, 175-191.
Fazel, S., & Danesh, J. (2002). Serious mental disorder in 23000 prisoners:
a systematic review of 62 surveys, Lancet, 359(9306), 545-50.
Field, A. (2013). Discovering Statistics Using IBM SPSS Statistics 4th
Edition. UK: London: SAGE Publications Ltd.
Fonagy, P. (1999). Male perpetrators of violence against women: An
attachment theory perspective, Journal of Applied Psychoanalytic
Studies, 1, 7-27.
Fruehwald, S., et al. (1998). The relevance of self-concepts discriminating
in long-term incarcerated sex offenders, Journal of Behavior
Therapy and Experimental Psychiatry 29(4), 267-278.
George, D., & Mallery, P. (2003). SPSS for Windows Step by Step: A Simple
Guide and Reference. 11.0 update (4th ed.). Boston: Allyn & Bacon.
61
Gillespie, W. (2005). Racial differences in violence and self-esteem among
prison inmates, American Journal of Criminal Justice 29(2), 161-
185.
Goldberg, B.R, Serper, M.R., Sheets, M., Beech, D., Dill, C., & Duffy, K.G.
(2007). Predictors of aggression on the psychiatric inpatient service:
Self-esteem, narcissism, and theory of mind deficits, The Journal of
Nervous and Mental Disease, 195(5), 436-442.
Gratz, K. L., Chapman, A. L., Dixon-Gordon, K. L., & Tull, M. T. (2016).
Exploring the association of deliberate self-harm with emotional
relief using a novel implicit association test, Personality Disorders:
Theory, Research, and Treatment, 7(1), 91-102.
Greenwald, A.G. & Banaji, M.R. (1995) Implicit social cognition: attitudes,
self-esteem, and stereotypes, Psychological Review, 102, 4–27.
Greenwald, A.G., & Farnham, S.D. (2000). Using the implicit association
test to measure self-esteem and implicit self-concept, Journal of
Personality & Social Psychology, 79(6), 1022-1038.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L. (1998). Measuring
individual differences in implicit cognition: The Implicit Association
Test. Journal of Personality and Social Psychology, 74, 1464-1480.
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding
and Using the Implicit Association Test: I. An Improved Scoring
Algorithm. Journal of Personality and Social Psychology, 85, 197-
216.
62
Gregg, A. P., & Sedikides, C. (2010). Narcissistic fragility: Rethinking its
links to explicit and implicit self-esteem, Self and Identity, 9, 142-
161.
Guerra, N.G., Huesmann, L.R., & Spindler, A. (2003). Community violence
exposure, social cognition, and aggression among urban elementary
school children, Child Development, 74(5), 1561-1576.
Gvion, Y., & Apter, A. (2011). Aggression, impulsivity and suicide
behaviour: a review of the literature, Archives of Suicide Research,
15(2), 93-112.
Hare, R. D. (2003). Manual for the Revised Psychopathy Checklist (2nd
ed.). Toronto, ON, Canada: Multi-Health Systems.
Hare, R. D., Clark, D., Grann, M., & Thornton, D. (2000). Psychopathy and
the predictive validity of the PCL-R: An international perspective,
Behavioral Sciences and the Law, 18(5), 623-645.
Hare, R.D., & Jutai, J.W. (1983). Criminal history of the male psychopath:
Some preliminary data. In: K.T. Van Dusen & S.A. Mednick (Eds).
Prospective Studies of Crime and Delinquency. pp. 225–236.
Kluwer-Nijhoff; Boston.
Hare, R.D. & McPherson, L.M. (1984). Violent and criminal behaviour by
criminal psychopaths. International Journal of Law and Psychiatry,
7, 33-50.
Harpur, T.J., Hare, R.D., & Hakistan, A.R. (1989). Two factor
conceptualization of psychopathy: construct validity and assessment
implications, Psychological Assessment: A Journal of Consulting &
Clinical Psychology, 1, 6-17.
63
Hayes, A. F. (2013). Introduction to Mediation, Moderation and
Conditional Process Analysis. New York, New York: Guilford
Press.
Hemphill, J. F., Hare, R. D., & Wong, S. (1998). Psychopathy and
recidivism: A review. Legal and Criminological Psychology, 3, 141-
172.
Hervé, H., Hayes, J. P., & Hare, R. D. (2003). Psychopathy and sensitivity
to the emotional polarity of metaphorical statements, Personality
and Individual Differences, 35(7), 1497-1507.
Hobson, J., Shine, J., & Roberts, R. (2000). How do Psychopaths Behave in
a Prison Therapeutic Community? Psychology, Crime & Law, 6(2),
139-154.
Hofmann, W., Gawronski, B., Gschwendner, T., Le., H., & Schmitt, M.
(2005). A meta-analysis on the correlation between the implicit
association test and explicit self-report measures, Personality &
Social Psychology Bulletin, 31(10), 1369-1385.
Howells, K. and Hollin, C. R. (1989) Clinical Approaches to Violence.
Chichester: John Wiley.
IBM Corp (2012). IBM SPSS Statistics for Windows, Version 21.0.
Armonk, NY: IBM Corp.
Institute for Economics & Peace (2013). The Economic Cost of Violence
Containment: A Comprehensive Assessment of the Global Cost of
Violence. Institute for Economics & Peace, Australia.
64
Johnstone, L., & Dallos, R. (2006). Formulation in Psychology and
Psychotherapy: Making Sense of People’s Problems. New York:
Routledge.
Jordan, C.H., Spencer, S.J., & Zanna, M.P. (2005). Types of high self-
esteem and prejudice: how implicit self-esteem relates to ethnic
discrimination among high explicit self-esteem individuals,
Personality and Social Psychology Bulletin, 31(5), 693-702.
Jordan, C. H., Spencer, S. J., Zanna, M. P., Hoshino-Browne, E., & Correll,
J. (2003). Secure and defensive high self-esteem, Journal of
Personality and Social Psychology, 85(5), 969-978.
Kantor, G.K., & Jasinski, J.L.(1998). Dynamics and risk factors in partner
violence. In: Jasinski JL, Williams LM, (Eds). Partner Violence: A
Comprehensive Review of 20 Years of Research. Thousand Oaks,
CA, Sage.
Kaplan, R.M., & Saccuzzo, D.P. (2009). Psychological Testing: Principles,
Applications & Issues. Wadsworth, Cengage Learning; USA.
Karpinski, A. (2004). Measuring self-esteem using the implicit association
test: The role of the other, Personality & Social Psychology Bulletin,
30(1), 22-34.
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(9), 539-46.
Kernberg, O. (1975). Borderline Conditions and Pathological Narcissism.,
New York: Jason Aronson Inc.
65
Kernberg O.F. (1986). Institutional problems of psychoanalytic education.
Journal of American Psychoanalytic Association, 4, 799-834.
Kernis, M.H. (1993). The role of stability and level of self-esteem in
psychological functioning. In R.F. Baumeister (Ed.), Self-esteem:
The puzzle of low self-regard (pp.167-182). New York: Plenum
Press.
Kernis, M. H. (2003). Toward a conceptualization of optimal self-esteem,
Psychological Inquiry, 14, 1-26.
Kernis, M.H. (2005). Measuring self-esteem in context: The importance of
stability of self-esteem in psychological functioning, Journal of
Personality, 73, 1–37.
Kernis, M. H., Grannemann, B. D. & Barclay, L. C. (1989). Stability and
level of self-esteem as predictors of anger arousal and hostility,
Journal of Personality and Social Psychology, 56, 1013-1023.
Kohut, H. (1966), "Forms and Transformations of Narcissism" in C. Strozier
(Ed.), Self Psychology and the Humanities, New York: Norton, 1985
pp. 97-123
Kohut, H. (1977). The Restoration of the Self. New York: International
Universities Press.
Kores Plesnicar, B., Zalar, B., Tomori, M., & Krajnc, I. (2003).
Measurement of simple reaction time in antipsychotic treatment of
patients with schizophrenia, Wiener klinische Wochenschrift, 115(1-
2), 58-62.
Krause, S., Back, M. D., Egloff, B., & Schmukle, S. C. (2011). Reliability
of implicit self-esteem measures revisited. European Journal of
66
Personality, 25, 239-251.
Krizan, Z., & Suls, J. (2009). Implicit self-esteem in the context of trait
models of personality, Personality and Individual Differences, 46,
659-663.
Langman, P. (2009). Why Kids Kill: Inside the Minds of School Shooters.
New York: Palgrave Macmillan.
Lansford JE, Dodge KA, Pettit GS, Bates JE, Crozier J, Kaplow J. (2002). A
12-year prospective study of the long-term effects of early child
physical maltreatment on psychological, behavioral, and academic
problems in adolescence. Archives of Pediatrics and Adolescent
Medicine, 156, 824–830.
Leary, M.R., & Tangney, J.P. (2005). The Handbook of Self & Identity, New
York: London: Guilford Press.
Levenson, M.; Kiehl, K.; Fitzpatrick, C. (1995). "Assessing psychopathic
attributes in a noninstitutionalized population". Journal of
Personality and Social Psychology, 68, 151-158.
Lima, E. N. (2007). The association between narcissism and implicit self-
esteem: A test of the fragile self-esteem hypothesis. (Doctoral
dissertation, Florida State University).
http://diginole.lib.fsu.edu/etd/1346/
Linehan, M.M. (1993). Skills Training Manual for Treating Borderline
Personality Disorder. NY: The Guildford Press.
Lowings, G.R. (2010). Improving Access to Neuropsychological Services,
unpublished report, Nottinghamshire Healthcare Trust,
Nottinghamshire.
67
Malle, B.F., & Horowitz, L.M. (1995). The puzzle of negative self-views:
An exploration using the schema concept, Journal of Personality &
Social Psychology, 68(3), 470-484.
McGuire, J. (2004). Minimising harm in violence risk assessment: Practical
solutions to ethical problems, Health, Risk and Society, 6, 327–345.
Meade, A. W. (2009). FreeIAT: An open-source program to administer the
implicit association test. Applied Psychological Measurement, 33,
643.
Miller, J.D., & Campbell, W.K. (2008). Comparing clinical and social-
personality conceptualizations of narcissism, Journal of Personality
76(3), 449-476.
Miller, J.D., Gaughan, E.T. & Pryor, L.R. (2008). The Levenson Self-
Report Psychopathy Scale: An examination of the personality traits
and disorders associated with the LSRP factors, Assessment, 15(4),
450-463.
Miller, J. D., Jones, S., & Lynam, D. R. (2011). Psychopathic traits from the
perspective of self- and informant-reports: Is there evidence for a
lack of insight? Journal of Abnormal Psychology, 120, 758-764.
Morrison, D., & Gilbert, P. (2001). Social rank, shame and anger in primary
and secondary psychopaths, The Journal of Forensic Psychiatry,
12(2), 330-356.
Nestor, P.G. (2002). Mental disorder and violence: personality dimensions
and clinical features, American Journal of Psychiatry, 159, 1973-
1978.
68
New, A.S. et al. (2004). Low prolactin response to fenfluramine in
impulsive aggression, Journal of Psychiatric Research, 38(3), 223-
230.
NHS England (2014). 2014/15 NHS standard contract for High secure
mental health services (Adults) Particulars, Schedule 2 – The
Services, A- Service Specifications. Service specification number:
C02/S/a
NHS Protect (2011). Tackling violent and anti-social behaviour in the NHS.
Joint working agreement between the Association of Police Officers
and NHS Protect. Accessed from:
https://www.cps.gov.uk/publications/agencies/joint_working_agree
ment.pdf on 28/03/16 at 15:51.
Office for National Statistics (2013). Statistical Bulletin: Crime in England
and Wales, Year Ending June 2013.
Office for National Statistics (2015). User Guide to Crime Statistics for
England & Wales. Newport; UK: Office for National Statistics.
Ostrowsky, M.K. (2010). Are violent people more likely to have low self-
esteem or high self-esteem? Aggression & Violent Behavior (15),
69-75.
Pincus, A.L., Ansell, E.B., Pimentel, C.A., Cain, N.M., Wright, A.G.C., &
Levy, K.N. (2009). Initial construction and validation of the
Pathological Narcissism Inventory, Psychological Assessment, 21,
365-379.
Raskin, R., & Hall, C. S. (1979). A narcissistic personality inventory,
Psychological Reports, 45, 590.
69
Raskin, R., Novacek, J., & Hogan, R. (1991a). Narcissism, self-esteem, and
defensive self-enhancement, Journal of Personality, 59(1), 19-38.
Raskin, R. & Terry, H. (1988). A Principle-Components Analysis of the
Narcissistic Personality Inventory and Further Evidence of Its
Construct Validity, Journal of Personality and Social Psychology,
54, 890-902.
Ray, J.V., Hall, J., Rivera-Hudson, N., Poythress, N.G., Lilienfeld, S.O., &
Morano, M. (2013). The relation between self-reported psychopathic
traits and distorted response styles: A meta-analytic review.
Personality Disorders: Theory, Research, and Treatment, 4, 1-14.
Reidy, D.R., Shelley-Tremblay, J.F., & Lilienfeld, S.O. (2011)
Psychopathy, Reactive Aggression, and Precarious Proclamations: A
Review of Behavioral, Cognitive, and Biological Research.
Aggression and Violent Behavior, 16, 512-524.
Rice, M. E., Harris, G. T., & Cormier, C. (1992). A follow- up of rapists
assessed in a maximum security psychiatric facility. Journal of
Interpersonal Violence, 5, 435–448.
Rosenberg, M. (1965). Society and the Adolescent Self-Image. Princeton,
NJ: Princeton University Press.
Sandstrom, M.J., & & Jordan, R. (2008). Defensive self-esteem and
aggression in childhood, Journal of Research in Personality, 42(2),
506-514.
Sarchiapone M, Carli V, Cuomo C, Marchetti M, & Roy, A. (2009).
Association between childhood trauma and aggression in male
prisoners, Psychiatry Research 165(1-2), 187-92.
70
Sarkar, J., & Duggan, C. (2010). Diagnosis and classification of personality
disorder: difficulties, their resolution and implications for practice,
Advances in Psychiatric Treatment, 16: 388-396.
Schröder-Abé, M., Rudolph, A. & Schütz, A. (2007). High implicit self-
esteem is not necessarily advantageous: Discrepancies between
explicit and implicit self-esteem and their relationship with anger
expression and psychological health. European Journal of
Personality, 21(3), 319-339.
Sedikides, C., & Gregg. A. P. (2003). Portraits of the self. In M. A. Hogg &
J. Cooper (Eds.), Sage handbook of social psychology (pp. 110–
138). London: Sage.
Sedikides, C., Rudich, E. A., Gregg, A. P., Kumashiro, M., & Rusbult, C.
(2004). Are normal narcissists psychologically healthy? Self-esteem
matters. Journal of Personality and Social Psychology, 87, 400–416.
Sellbom, M. (2011). Elaborating on the Construct Validity of the Levenson
Self-Report Psychopathy Scale in Incarcerated and Non-Incarcerated
Samples. Law and Human Behavior, 35, 440–451.
Seto, M. C., & Barbaree, H. (1999). Psychopathy, treatment behavior, and
sex offender recidivism. Journal of Interpersonal Violence, 14,
1235–1248.
Shanock, L. R., Baran, B. E., Gentry, W. A., Pattison, S. C., & Heggestad,
E. D. (2010). Polynomial regression with response surface analysis:
A powerful approach for examining moderation and overcoming
limitations of difference scores. Journal of Business and Psychology,
25(4), 543-554.
71
Short, L.A., et al., (2012). Detection of propensity for aggression based on
facial structure irrespective of face race, Evolution & Human
Behaviour, 33(2), 121-129.
Skeem, J. L., Poythress, N., Edens, J. F., Lilienfeld, S. O.,& Cale, E. M.
(2002). Psychopathic personality or personalities? Exploring
potential variants of psychopathy and their implications for risk
assessment. Aggression and Violent Behavior, 8, 513–546.
Singer, M. I., Anglin, T. M., Song, L., & Lunghofer, L. (1995).
Adolescents’ exposure to violence and associated symptoms of
psychological trauma, JAMA, 273, 477-482.
Snowden, R., Gray, N., Smith, J., Morris, M., & MacCulloch, M. (2004).
Implicit affective associations to violence in psychopathic
murderers, The Journal of Forensic Psychiatry & Psychology, 15(4),
620-641.
StataCorp. (2015). Stata Statistical Software: Release 14. College Station,
TX: StataCorp LP.
Steele C., & Aronson J. (1995). Stereotype threat and the intellectual test
performance of African Americans, Journal of Personality and
Social Psychology, 69, 797–811.
Stoessel, B. J. (2008). Facets of the self in grandiose and vulnerable
narcissism: implicit self-esteem, explicit self-esteem, and shame.
Dissertation Abstracts International, 68(7). (UMI 3275751)
Svindseth, M. F., et al. (2008). Narcissism in patients admitted to
psychiatric acute wards: Its relation to violence, suicidality and other
psychopathology, BMC Psychiatry, 8, 1.
72
Tabachnick, G. G., and Fidell, L. S. (2007). Experimental Designs Using
ANOVA. Belmont, CA: Duxbury.
Tafarodi, R. W., & Ho, C. (2006). Implicit and explicit self-esteem: What
are we measuring? Canadian Psychology, 47 (3), 195-202.
Tan, L. & Grace, R.C. (2008). Social desirability and sexual offenders: A
review, Sexual Abuse: A Journal of Research and Treatment, 20 (1),
61-87.
Tew, J., Harkins, L., & Dixon, L. (2013). What works in reducing violent
re-offending in psychopathic offenders, in L.A Craig, L. Dixon &
T.A Gannon (Eds) “What Works in Offender Rehabilitation: An
Evidence-Based Approach to Assessment and Treatment”, John
Wiley & Sons, p.129-141.
Toch, H., Adams, K., & Grant, D. (1989). Coping: Maladaptation in
Prisons. New Brunswick, NJ: Transaction.
Vater, A., Ritter, K., Schröder-Abé, M., Schütz, A., Lammers, C-H.,
Bosson, J.K., & Roepke, S. (2013). When grandiosity and
vulnerability collide: Implicit and explicit self-esteem in patients
with narcissistic personality disorder, Journal of Behavior Therapy
& Experimental Psychiatry, 44, 37-47.
Verona, E., Patrick, C. J., & Joiner, T. T. (2001). Psychopathy, antisocial
personality, and suicide risk. Journal of Abnormal Psychology, 104,
462-470.
Walker, S. (2005). Gender differences in the relationship between young
children’s peer-related social competence and individual differences
73
in theory of mind. The Journal of Genetic Psychology, 166(3), 297-
312.
Walker, J.S., & Bright, J.A. (2009). False inflated self-esteem and violence:
a systematic review and cognitive model, The Journal of Forensic
Psychiatry & Psychology, 20(1), 1-32.
Walker JS, Gudjonsson GH (2006). The Maudsley Violence Questionnaire:
Relationship to personality and self-reported offending, Personality
and Individual Differences, 40, 795-806.
Walker, J., & Knauer, V. (2011). Humiliation, self-esteem and violence, The
Journal of Forensic Psychiatry & Psychology, 22(5), 724-741.
Washburn, J.J., McMahon, S.D., King, C.A., Reinecke, M.A. & Silber, C.
(2004). Narcissistic features in young adolescents: Relations to
aggression and internalising symptoms, Journal of Youth &
Adolescence, 33, 247-260.
Webster, C.D., Douglas, K.S., Eaves, S.D., & Hart, S.D. (1997a). Assessing
risk of violence to others. In C.D. Webster & M.A. Jackson (Eds.),
Impulsivity: Theory, Assessment, and Treatment (pp. 251-277). New
York: Guilford.
Williams, H.W., Cordan, G. , Mews, A.J. , Tonks, J. and Burgess, N.W.
(2010). Self-reported traumatic brain injury in male young offenders:
a risk factor for re-offending, poor mental health and violence,
Neuropsychological Rehabilitation, 20(12), 801-12.
Williamson S., Hare R. D., & Wong S. (1987). Violence: criminal
psychopaths and their victims, Canadian Journal of Behavioural
Science, (19), 454-462.
74
Wong, S. (1988). Is Hare’s Psychopathy Checklist reliable without the
interview? Psychological Reports, 62, 931–934.
Woodworth, M., & Porter, S. (2002). In cold blood: Characteristics of
criminal homicides as a function of psychopathy. Journal of
Abnormal Psychology, 111, 436-445.
Young, J.E., Klosko, J.S., & Weishaar, M.E. (2006). Schema Therapy: A
Practitioner's Guide. Guilford Press.
Zeigler-Hill, V. (2011). The connections between self-esteem and
psychopathology, Journal of Contemporary Psychotherapy, 41, 157-
164.
Zeigler-Hill, V., & Jordan, C.H. (2011). Behind the mask: Narcissism and
implicit self-esteem. In Campbell, W.K., & Miller, J.D. (Eds). The
Handbook of Narcissism and Narcissistic Personality Disorder:
Theoretical Approaches, Empirical Findings, and Treatments, UK:
John Wiley & Sons.
75
MRP Empirical Paper Appendices
Appendix 1
From the journal website: http://www.guilford.com/journals/Journal-of-Personality-Disorders/Krueger-Livesley/0885579X
Journal of Personality Disorders
Instructions to Authors
Types of Articles
Regular Articles: Reports of original work should not normally exceed 30 pages (typed, double-lined spaces, and with standard margins, including tables, figures, and references). Occasionally, an author may feel that he or she needs to exceed this length (e.g., a report of a series of studies, or a report that would benefit from more extensive technical detail). In these circumstances, an author may submit a lengthier manuscript, but the author should describe the rationale for a submission exceeding 30 pages in the cover letter accompanying the submission. This rationale will be taken into account by the Editors, as part of the review process, in determining if the increased length is justified.
Invited Essays and Special Articles: These articles provide an overview of broad-ranging areas of research and conceptual formulations dealing with substantive theoretical issues. Reports of large-scale definitive empirical studies may also be submitted. Articles should not exceed 40 pages including tables, figures, and references. Authors contemplating such an article are advised to contact the editor in advance to see whether the topic is appropriate and whether other articles in this topic are planned.
Brief Reports: Short descriptions of empirical studies not exceeding 20 pages in length including tables, figures, and references.
Web-Based Submissions: Manuscripts must be produced electronically using word processing software, double spaced, and submitted along with a cover letter to http://jpd.msubmit.net. Authors may choose blind or non-blind review. Please specify which option you are choosing in your cover letter. If you choose blind review, please prepare the manuscript accordingly (e.g., remove identifying information from the first page of the manuscript, etc.). All articles should be prepared in accordance with the Publication Manual of the American Psychological Association. They must be preceded by a brief abstract and adhere to APA referencing format.
Tables should be submitted in Excel. Tables formatted in Microsoft Word’s Table function are also acceptable. (Tables should not be submitted using tabs, returns, or spaces as formatting tools.)
Figures must be submitted separately as graphic files (in order of preference: tif, eps, jpg, bmp, gif; note that PowerPoint is not acceptable) in the highest possible resolution. Figure caption text should be included in the article’s Microsoft Word file. All figures must be readable in black and white.
76
Permissions: Contributors are responsible for obtaining permission from copyright owners if they use an illustration, table, or lengthy quote (100+ words) that has been published elsewhere. Contributors should write both the publisher and author of such material, requesting nonexclusive world rights in all languages for use in the article and in all future editions of it.
References: Authors should consult the publication manual of the American Psychological Association for rules on format and style. All research papers submitted to the Journal of Personality Disorders must conform to the ethical standards of the American Psychological Association. Articles should be written in nonsexist language. Any manuscripts with references that are incorrectly formatted will be returned by the publisher for revision.
Sample References: Davis, C. G., & McKearney, J. M. (2003). How do people grow from their experience with trauma or loss? Journal of Social & Clinical Psychology, 22(5), 477-492.
Dweck, C., & Wortman, C. (1982). Learned helplessness, anxiety and achievement. In H. Kron & L. Laux (Eds.), Achievement, stress, and anxiety (pp. 93-125). Washington, DC: Hemisphere Publishing Group.
Roelofs, J., Meesters, C., Ter Huurne, M., Bamelis, L., & Muris, P. (2006). On the links between attachment style, parental rearing behaviors, and internalizing and externalizing problems in nonclinical children. Journal of Child and Family Studies, 15, 331-344.
77
Appendix 2
Methods of Measuring Fragile Self-Esteem
There are a number of ways of capturing fragile self-esteem, which will be
outlined below along with a discussion of their strengths and weaknesses.
Difference Scores
For example some researchers have measured fragile self-esteem by
computing the difference between standardised ISE and ESE scores, with a
higher score indicating a larger discrepancy. The researchers also created a
dummy variable to capture the direction of the discrepancy (i.e. ISE>ESE or
ISE<ESE, Creemers et al., 2013). By analysing the interaction between the
size of and direction of the discrepancy, Creemers et al (2013) found that
high implicit/low explicit SE (known as damaged self-esteem in the
literature) was associated with depression, suicidal ideation and loneliness,
and that high explicit/low implicit SE was associated with loneliness.
However the use of difference scores for analysing discrepancies has been
strongly criticised in the literature and a number of associated
methodological problems have been highlighted (Edwards, 2001; Edwards,
2002; Kaplan & Saccuzzo, 2009; Shanock et al., 2010), including
specifically in research on self-related concepts (Cafri, van den Berg, &
Brannick, 2010). Using difference scores may lead to more easily
interpretable findings however by using difference scores only, the relative
influence of each original variable is discounted.
Interaction Effects With Simple Slopes Tests
An alternative method of measuring fragile self-esteem would be to conduct
a regression analyses, following up significant interactions between ESE
and ISE with simple slopes tests. This would include the individual effect
78
of both ESE and ISE, as well as the interaction between these two variables.
This method allows exploration of whether the interaction effect provides
any additional explanatory power over and above their separate effects of
ESE and ISE. A further issue that that by using a difference score (one
variable minus another which would pool the error from the two variables)
information from the individual variables, including measurement error,
would be lost from statistical modelling. The moderated regression method
performed, by including the two variables plus their interaction, would
retain more information for statistical analysis. Using the interaction term
followed by simple slopes analysis to explore the direction of the interaction
is a more valid method of capturing fragile self-esteem as defined by the
Partial Discrepancy Model of fragile self-esteem (Gregg & Sedikides,
2010). This method has been well validated by other researchers and is the
most common method used in research in this area (Gregg & Sedikides,
2010; Schroder-Abe et al., 2007; Borton et al., 2012; Zeigler-Hill, 2006).
79
Appendix 3
Details of FreeIAT software
“FreeIAT” is a free, open source version of the Implicit Association Test from the following website:
http://www4.ncsu.edu/~awmeade/FreeIAT/FreeIAT.htm
“The FreeIAT is an administration program that assesses reaction time and whether the response is correct or incorrect for different stimuli. The version used in the present study contained text only.
The first stage begins and stimuli are randomly selected for administration with the constraint that the same stimulus may not be presented twice in a row (else the effect is the perception that the program did not record the response as the screen does not noticeably change). A large X appears if the respondent answers incorrectly and the screen does not change until the correct response is chosen.
For each trial in all stages, the stimulus name, whether or not the respondent answered correctly, and the response time (in milliseconds) is recorded. Note that the time stored is the time until a correct response is given, thus incorrect responses do not "stop the clock". A counter tracks how many trials have been administered in each stage. When the maximum number of trials has been reached for a given stage, the program clears the screen and displays the relevant category labels for the next stage's stimuli so that respondents may examine them before beginning the next stage.
Stages 3 and 5 involve paired comparisons. For these stages, the type of stimulus to be administered is first randomly chosen, then the stimulus within that type is randomly chosen. As a result, the number of stimuli in each type can differ (e.g., 4 of each type of image may be used along with 10 of positive words and 10 negative words) without displaying a disproportionate number of words to images.
During each trial, information is stored regarding the current stage, what stimulus was administered, how long (in milliseconds) until a correct response was entered, and whether an incorrect response was given. This information is used to compute scores and is reported in an output file.”
Instructions for the test:
“You will be presented with a set of words or images to classify into groups using the “e” and “I” keys on the keyboard. Classify items as quickly as you can while making as few mistakes as possible. Going too slow or making too many mistakes will result in an interpretable score.”
80
Table 1. List of words from the original IAT (Gregg & Sedikides, 2010):Self Non-Self Pleasant UnpleasantMeMyselfI
TheyThemThose
ExcellentHeavenJoyTrustPeaceEnjoymentFriendHonestSweetheartLoveFreedomParadise
MurderCancerWarDisasterHatredSlaughterBombAgonyTortureSlimeFilthTraitor
The word “bomb” was replaced with a word of similar valance, arousal and dominance following feedback from patients that this word may be particularly distressing for patients experiencing paranoia and/or from a Muslim background. The word “murder” was also replaced with a word of similar valence, arousal and dominance because participants were likely to strongly associate “self” words to this word due to their forensic history rather than because of low-implicit self-esteem. Replacement words were obtained from the Affective Norms for English Words (ANEW; Bradley & Lang, 2010), ensuring that there was no significant differences between valance, dominance and arousal ratings (see Appendix 2).
Table 2. List of words used in the revised taskSelf Non-Self Pleasant UnpleasantMeMyselfI
TheyThemThose
ExcellentHeavenJoyTrustPeaceEnjoymentFriendHonestSweetheartLoveFreedomParadise
TerrifiedCancerWarDisasterHatredSlaughterStressAgonyTortureSlimeFilthTraitor
The number of pleasant (12), unpleasant (12), self (3), and non-self (3) words is the same in the original and the revised IAT.
Below are some screenshots of the FreeIAT used in the present study:
81
Figure 1. Screenshot From Introduction Screen of IAT
Figure 2. Screenshot From Block 1 of the IAT
82
Figure 3. Screenshot From Block 3 of the IAT
Table 3Sequence of trial blocks in the IAT used in the present studyBlock No. of
TrialsFunction Items assigned
to left key response
Items assigned to right key response
1 20 Practice Self Non-self2 20 Practice Pleasant Unpleasant3 40 Test Pleasant + self Unpleasant + non-
self4 20 Practice Non-self Self5 40 Test Pleasant + non-
selfUnpleasant + self
83
Appendix 4
T Tests Comparing IAT Words From Gregg and Sedikides (2010) IAT and Replacement IAT Words to be Used in the Present Study
Independent samples t-tests were conducted to compare valence, arousal and
dominance scores for words in the original and revised IAT.
Table 4.Independent samples t test comparing valance ratings of original and revised IAT words
Levene’s Test for
Equality of Variances
T-Test for Equality of Means
F Sig. t df Sig. (2-tailed)
Mean Difference
Std. Error Difference
95% Confidence Interval of the Difference
Lower UpperEqual variances assumed
.007 .933 .129 22 .899 .0333 .259 -.505 .571
Equal variances not assumed
- - .129 22 .899 .0333 .259 -.505 .571
There was no significant difference in scores for the original (M=2.46,
SD=0.64) and the revised IAT (M=2.42, SD=0.63, t(22)=.129, p=.90) in
terms of valence (see Table 4).
Table 5.Independent samples t test comparing arousal ratings of original and revised IAT words
Levene’s Test for
Equality of Variances
T-Test for Equality of Means
F Sig. t df Sig. (2-tailed)
Mean Difference
Std. Error Difference
95% Confidence Interval of the Difference
Lower
Upper
Equal variances assumed
.090
.766
-.152
22 .881 -.071 .467 -1.04 .90
84
Equal variances not assumed
- - -.152
21.928 .881 -.071 .467 -1.04 .90
There was no significant difference in scores for the original (M=6.08,
SD=1.11) and the revised IAT (M=6.15, SD=1.18, t(22)=-.15, p=.88, see
Table 5).
Table 6.Independent samples t test comparing dominance ratings of original and revised IAT words
Levene’s Test for
Equality of Variances
T-Test for Equality of Means
F Sig. t df Sig. (2-tailed)
Mean Difference
Std. Error Difference
95% Confidence Interval of the Difference
Lower
Upper
Equal variances assumed
.004
.947
.424
22 .676 .16 .377 -.623 .943
Equal variances not assumed
.424
21.965
.676 .16 .377 -.623 .943
There was no significant difference in scores for the original (M=4.55,
SD=.94) and the revised IAT (M=4.39, SD=.91, t(22)=.42, p=.68, see Table
6). The independent-samples t-tests indicate that there are no significant
differences between valence, arousal and dominance ratings of the words
from the original and updated IAT task.
85
Appendix 5
Methods of scoring the IAT
Table 7.A comparison of IAT scoring methodsConventional AlgorithmGreenwald, McGhee and Schwartz (1998)
Revised AlgorithmGreenwald, Nosek and Banaji (2003)
FreeIAT Algorithm
Method used in this study
7 blocks (use data from B4 and 7 only – these are the two critical blocks)
No elimination of excessively slow responding and/or high error rates. Drops first two trials of each block.
Extreme value treatment: Recodes latencies outside 300/3000 ms boundaries to that boundary value
7 blocks (use data from B3, 4, 6 and 7 – two critical plus 2 practice blocks preceding them)
GNB recommend keeping all respondents
Extreme value treatment: eliminates trials with latencies > 10,000 ms and eliminates participants with more than 10% of trials with latencies of less than 300 ms. Recommends deleting trials less than 400 ms
5 blocks. Whilst there are usually 7, Stages 3 and 4 are identical (and both are scored) as are stages 6 and 7. Thus the FreeIAT marries these stages though the resulting score is nearly identical to that suggested by Greenwald et al
Uses all trials/keeps all respondents
No extreme value treatment (GNB say you should keep in all respondents unless they score 300 ms for more than 10% of their trials).
I have collected data over 5 blocks using the FreeIAT program
Use all trials/keep all respondents
Eliminate respondents who score 300 ms for more than 10% of their trials (none in my data set). Recoded latencies outside 150/5000 ms boundaries to that boundary value as there are some very long latencies in my data set (this is what Gregg & Sedikides 2010 did)
Conventional AlgorithmGreenwald, McGhee and
Revised AlgorithmGreenwald, Nosek and Banaji
FreeIAT Algorithm
Method Used in this Study
86
Schwartz (1998) (2003)Kept in error responses in average latency score
IAT score computation: Averages latencies for B4 and B7. Computes the difference B7 – B4
Log transformed resulting values
Replaces error latencies with block mean + 600 ms
IAT score computation: Computes the mean RT for items in Block 3 in which the initial response was correct. Computes the mean RT for items in Block 5 in which the initial response was correct. Computes one pooled SD for all trials in B3 and 6, another for B4 and 7
No log transformation. They found that the D measure was better than the log transformed measure where they used SD as a divisor to adjust difference between means for this effect of underlying variability.
Replaces error latencies with block mean + 600 ms
IAT score computation: Average corrected reaction time for B5 minus average corrected RT for B3 minus pooled SDof all items (regardless of whether they were initially responded to in a correct or incorrect manner) see GNB, 2003, p. 201, 2nd paragraph. Note, formula uses N, not N-1, in denominator. This is what GNB call the pooled SD (see number 5 in scoring key).
No log transformation (see comment in previous column)
Replaces error latencies with block mean + 600 ms
IAT score computation: Average corrected reaction time for B5 minus average corrected RT for B3 minus pooled SDof all items (Free IAT version)
No log transformation (see comment in previous column)
87
Appendix 6
Ethical Approval Documents
Letter Following REC Meeting
88
89
90
91
Letter following submission of further information
92
93
94
95
R&D Approval Letter
96
97
Substantial Amendment Approval
98
99
100
R & D Approval Email for Amendment
101
Ethical Approval Letter from The University of Surrey
Letter of Sponsorship from the University of Surrey
102
103
104
Appendix 7
Normal Distribution Assumptions Testing
Table 8.Descriptive statistics including skewness & kurtosis values
N Minimum Maximum Mean Std. Deviation Skewness Kurtosis
Statistic Statistic Statistic Statistic Statistic Statistic Std. Error Statistic Std. Error
Explicit SE (RSES) 50 3 30 19.54 6.062 -.406 .337 -.003 .662SRPS Factor 1 49 17 53 32.63 7.008 .109 .340 .309 .668SRPS Factor 2 49 12 34 23.63 4.889 -.175 .340 .094 .668SRPS Total 49 39 79 56.27 10.547 -.134 .340 -.809 .668Narcissism Total (NPI) 48 0 29 12.75 7.201 .409 .343 -.303 .674Adaptive Narcissism 48 0 12 5.58 3.407 .262 .343 -.867 .674Maladaptive Narcissism 48 0 13 4.50 3.032 .722 .343 .513 .674Counts of physically aggressive convictions
50 1 28 7.14 6.725 1.532 .337 1.811 .662
Severity of Index Offence (MOAS)
50 0 26 15.60 5.525 -1.352 .337 2.589 .662
Implicit SE (IAT) 50 -.589 1.174 .461 .383 -.367 .337 .257 .662Number of Incidents 50 0 12 .74 2.0 4.372 .337 21.68 .662PCLR Factor 1 44 1 16 7.93 3.566 .255 .357 -.403 .702PCLR Factor 2 44 0 20 12.52 5.036 -.744 .357 -.231 .702PCLR Total 50 4 36 22.21 7.814 -.559 .337 -.152 .662
105
Note: SE=self-esteem, RSES=Rosenberg Self-Esteem Scale, SRPS=Self-Report Psychopathy Scale, NPI= Narcissistic Personality Inventory, MOAS=Modified Overt Aggression Scale, IAT=Implicit Association Test, PCL-R=Psychopathy Checklist-Revised (clinician-rated psychopathy).
106
Tests of NormalityCase Processing Summary
Cases
Valid Missing Total
N Percent N Percent N Percent
PhysAggresConv_Incl_Robbery 50 100.0% 0 0.0% 50 100.0%
Incident_count_12m 50 100.0% 0 0.0% 50 100.0%
SRPS_F1 49 98.0% 1 2.0% 50 100.0%
SRPS_F2 49 98.0% 1 2.0% 50 100.0%
SRPS_Tot 49 98.0% 1 2.0% 50 100.0%
NPI_adapt 48 96.0% 2 4.0% 50 100.0%
NPI_maladapt 48 96.0% 2 4.0% 50 100.0%
NPI_Total 48 96.0% 2 4.0% 50 100.0%
PCLR_F1 44 88.0% 6 12.0% 50 100.0%
PCLR_F2 44 88.0% 6 12.0% 50 100.0%
PCLR_Total 50 100.0% 0 0.0% 50 100.0%
RSES 50 100.0% 0 0.0% 50 100.0%
IAT_Tot_Score 50 100.0% 0 0.0% 50 100.0%
MoasIndSevere_Linear 50 100.0% 0 0.0% 50 100.0%
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
Statistic df Sig. Statistic df Sig.
Age .064 50 .200* .969 50 .217
PhysAggresConv_Incl_Robbery .247 50 .000 .811 50 .000
Incident_count_12m .356 50 .000 .411 50 .000
SRPS_F1 .088 49 .200* .978 49 .477
SRPS_F2 .122 49 .066 .980 49 .546
SRPS_Tot .116 49 .095 .955 49 .056
NPI_adapt .116 48 .105 .957 48 .076
NPI_maladapt .143 48 .016 .948 48 .032
NPI_Total .090 48 .200* .971 48 .278
PCLR_F1 .092 44 .200* .981 44 .689
PCLR_F2 .141 44 .029 .935 44 .016
PCLR_Total .129 50 .036 .962 50 .103
RSES .060 50 .200* .980 50 .551
IAT_Tot_Score .085 50 .200* .984 50 .706
MoasIndSevere_Linear .341 50 .000 .819 50 .000
*. This is a lower bound of the true significance.
a. Lilliefors Significance Correction
109
Physically Aggressive Convictions Including Robbery
111
Counts of Incidents in the Past 12 Months
112
113
Self-Report Psychopathy Scale Factor 1
114
115
Self-Report Psychopathy Scale Factor 2
116
117
Self-Report Psychopathy Scale Total
118
119
Adaptive Subscale of the Narcissistic Personality Inventory (NPI)
120
121
Maladaptive Subscale of the Narcissistic Personality Inventory (NPI)
122
123
Narcissistic Personality Inventory (NPI) Total Score
124
125
Psychopathy Checklist Factor 1
126
127
Psychopathy Checklist Factor 2
128
129
Psychopathy Checklist Total Score
130
131
Rosenberg Self-Esteem Scale (Explicit Self-Esteem)
132
133
Implicit Association Test D Score (Implicit Self-Esteem)
134
135
Severity Ratings of Index Offence (MOAS Scores)
136
137
Appendix 8
Frequency Tables for Categorized Variables
Table 9.Frequency table for incidents in the Past 12 Months (categorized into two groups)
Frequency Percent Valid Percent Cumulative Percent
0 35 70.0 70.0 70.0
1 15 30.0 30.0 100.0
Total 50 100.0 100.0
Note: 0=no incidents in the past 12 months, 1=at least one incident in the past 12 months
Table 10.Severity ratings of index offence (categorized into 3 groups)
Frequency Percent Valid Percent Cumulative Percent
0 10 20.0 20.0 20.0
1 23 46.0 46.0 66.0
2 17 34.0 34.0 100.0
Total 50 100.0 100.0
Note: 0= score of less than 16 on the MOAS, 1= score of 16 on the MOAS, 2= score of greater than 16 on the MOAS
Table 11.Index offence characterized by reactive or instrumental aggression from CGVI (categorized into 2 groups)
Frequency Percent Valid Percent Cumulative Percent
0 19 38.0 40.4 40.4
1 28 56.0 59.6 100.0
Sub-Total
47 94.0 100.0
Missing 3 6.0 Total 50 100.0
Note: 0= primarily reactive, 1= primarily instrumental. CGVI=Coding Guide for Violent Incidents
138
Appendix 9
Extra Regression Analyses Exploring Self-Esteem Fragility and Aggression Controlling for Narcissism
Does Self-esteem fragility predict higher levels of aggression, controlling
for narcissism?
Physically Aggressive Convictions Including Robbery
To explore whether fragile self-esteem would predict a higher number of physically aggressive convictions controlling for narcissism, a zero-truncated negative binominal regression was used with number of physically aggressive convictions including robbery as the dependent variable and with explicit self-esteem (RSES scores), implicit SE (IAT score), the interaction between these two variables and narcissism (NPI scores) as continuous predictor variables (see Table 12). The model as a whole was not statistically significant (LR χ2(4, N = 48) = 2.4, p =.663). The model as a whole explained 0.9% (Pseudo R2= .0086) of the variance in counts of convictions. None of the individual variables made a significant contribution to the model. These results do not support the hypothesis that discrepant self-esteem would predict a higher number of physically aggressive convictions (including robbery) controlling for narcissism. The alpha dispersion parameter (alpha= .1.065) was significantly greater than zero (likelihood ratio χ2 test that alpha equals 0 (df=1)= 136.95), which confirms that the data are over dispersed and a negative binomial model provides a better estimation of the data than a zero truncated poisson model.
Table 12. Regression model predicting number of physically aggressive convictions excluding robbery, controlling for narcissism
Coef S.Error
z p 95% CI interval for Exp(B)Lower
Upper
RSES -.253 .197 -1.29
.199
-.640 .133
IAT .123 .172 .72 .473
-.214 .460
Interaction
.108 .255 .42 .671
-.391 .607
NPI .156 .199 .78 .435
-.235 .546
Constant 1.781
.186 9.56
.000
1.416 2.146
Alpha 1.065
.438 - - .473 2.383
139
LR test of alpha=0: chibar2(01) = 136.95 Prob >= chibar2 = 0.000
Incidents on the Ward
To test the hypothesis that fragile self-esteem would predict higher levels of
institutional aggression controlling for narcissism, a logistic regression was
performed with group membership (no incidents in the past 12 months or 1
or more incidents in the past 12 months) as the binary dependent variable
and explicit self-esteem (RSES scores), implicit SE (IAT score), the
interaction between these two variables (RSES*IAT) and narcissism (NPI
scores) as continuous predictor variables. The full model was statistically
significant, (χ2(48, 4)=.9.826, p = .043), indicating that the model was able
to predict group membership. The model as a whole explained between
18.5% (Cox and Snell R Square=.185) and 26.4% (Nagelkerke R
square= .264) of the variance in group membership, and correctly classified
79.2% of the cases, which is better than the model’s predictive ability with
only the constant in the equation (70.8%). Individually, narcissism
(Beta=1.050, p<.05) and explicit self-esteem (Beta=-.995, p<.05) made
significant unique contributions to the model (see Table 13). So high
narcissism and low explicit self-esteem appear to be associated with
institutional aggression.
Table 13. Regression model predicting institutional aggression, controlling for narcissism
Beta S.Error
Wald
P Exp(B)
95% CI interval for Exp(B)Lower
Upper
RSES -.995 .428 5.399
.020
.370 .160 .856
IAT .318 .405 .617 .432
1.374 .622 3.037
140
Interaction
.269 .445 .366 .545
1.309 .547 3.130
NPI 1.050
.436 5.783
.016
2.856 1.214 6.719
Constant -1.143
.395 8.385
.004
.319 - -
Severity of Aggression in Index Offence
To explore whether fragile self-esteem will predict higher severity of
aggression in the index offence, controlling for narcissism an ordinal
regression analysis was performed to determine whether the odds of
committing a more aggressive index offence differed significantly according
to levels of explicit, implicit, discrepant (the explicit*implicit interaction)
self-esteem and narcissism. Group membership (index offence MOAS score
lower than 16, MOAS score of 16 or MOAS score of greater than 16) was
the ordinal dependent variable and explicit self-esteem (RSES scores),
implicit self-esteem (IAT score), the interaction (representing discrepant
self-esteem) and narcissism (NPI scores) were the continuous predictor
variables (see Table 14). The test of parallel lines was not significant (χ2(4)
= 2.111, p = .715) indicating that each independent variable has an identical
effect at each cumulative split of the ordinal dependent variable, thus the
assumption of proportional odds has been met. The full model containing all
predictor variables was not a significantly better fit than the model without
the predictor variable (χ2(4) = 7.104, p = .130). The model as a whole only
explained 15.7% (Nagelkerke Pseudo R square= .157) of the variance in
group membership. Findings indicate that for a one unit increase in explicit
self-esteem, we expect a .085 increase in the ordered log odds of being in a
141
higher severity of index offence group, however this finding was not
significant (p=.786). For a one unit increase in implicit self-esteem, we
expect a .444 decrease in the ordered log odds of being in a higher severity
of index offence group, however this finding was not significant (p=.162).
For a one unit increase in narcissism, we expect a .272 increase in the
ordered log odds of being in a higher severity of index offence group,
however this finding was not significant (p=.385).
Table 14. Regression model predicting severity of index offence
Estimate
Std.Error
Wald
p 95% CI
Lower
Upper
Threshold
MOAS Group 0
-1.666 .402 17.166
.000
-2.454
-.878
MOAS Group 1
.636 .319 3.974
.046
.011 1.262
Location
RSES .085 .312 .074 .786
-.527
.696
IAT -.444 .317 1.955
.162
-1.066
.178
RSES*IAT Interaction
.861 .456 3.571
.059
-.032
1.755
NPI .272 .313 .756 .385
-.341
.885
Instrumental or Reactive Aggression
To explore whether self-esteem fragility will predict whether the index
offence involved more instrumental or reactive aggression controlling for
narcissism, a logistic regression was performed with group membership
(instrumental or reactive) as the binary dependent variable and explicit self-
esteem (RSES scores), implicit SE (IAT score), the interaction between
142
these two variables (RSES*IAT) and narcissism (NPI scores) as continuous
predictor variables (see Table 13). The full model was not statistically
significant, (χ2(45, 4)=.3.416, p = .491), indicating that the model was not
able to predict group membership. The model as a whole explained between
7.3% (Cox & Snell R Square = .073) and 10% (Nagelkerke R Square
= .100) of the variance in group membership, and correctly classified 60%
of the cases, which is less than the model’s predictive ability with only the
constant in the equation 62.2%). Individually, none of the predictor
variables contributed significantly to the model (see Table X). This indicates
that there was no support for the hypothesis that self-esteem fragility
predicts whether the index offence involved more instrumental or reactive
aggression controlling for narcissism.
Table 15. Regression model predicting reactive or instrumental aggression, controlling for narcissism
Beta
S.Error
Wald
p Exp(B)
95% CI interval for Exp(B)Lower
Upper
RSES .247 .348 .503 .478
1.280 .647 2.532
IAT -.555
.379 2.143
.143
.574 .273 1.207
Interaction
-.095
.434 .048 .826
.909 .388 2.130
NPI -.343
.359 .914 .339
.709 .351 1.434
Constant .593 .332 3.195
.074
1.810 - -
Comparison of model with and without narcissism in it: Subtracting the R-
Square value for the model including narcissism (Cox & Snell R Square
= .073) and 10% (Nagelkerke R Square = .100) as a predictor from the
model that does not include it (Cox and Snell R Square=.018) to 2.4%
143
(Nagelkerke R square= .024)). So .073-.018= .055 and .100 - .024= .076. So
the model with PCL-R in it explains between 5.5% and 7.6% of the
additional variance in institutional aggression grouping. This indicates that
there was no support for the hypothesis that discrepant self-esteem would
predict higher levels of aggression controlling for narcissism.
Appendix 10
ESE, ISE, interaction and psychopathy regressions
Does Self-esteem fragility predict higher levels of aggression, controlling
for psychopathy?
Physically Aggressive Convictions
To explore whether discrepant self-esteem would predict a higher number of
physically aggressive convictions controlling for psychopathy, a zero-
truncated negative binominal regression was used with number of physically
aggressive convictions including robbery as the dependent variable and with
explicit self-esteem (RSES scores), implicit SE (IAT score), the interaction
between these two variables and clinician-rated psychopathy (PCL-R
scores) as continuous predictor variables (see Table 16). The model as a
whole was statistically significant (LR χ2(4, N = 50) = 11.96, p = .017). The
model as a whole explained 4.1% (Pseudo R2= .041) of the variance in
counts of convictions. Only one individual variable (PCL-R score) made a
significant contribution to the model (β =.571, p=.001). These results do not
support the hypothesis that discrepant self-esteem would predict a higher
144
number of physically aggressive convictions (including robbery) controlling
for psychopathy, because psychopathy was the only significant predictor.
The alpha dispersion parameter (alpha= .717) was significantly greater than
zero (likelihood ratio χ2 test that alpha equals 0 (df=1) = 105.24), which
confirms that the data are over dispersed and a negative binomial model
provides a better estimation of the data than a zero truncated poisson model.
Table 16. Regression model predicting number of physically aggressive convictions including robbery
Coef S.Error
z p 95% CI interval for Exp(B)Lower
Upper
RSES -.141 .153 -.92 .360
-.441 .160
IAT .133 .140 .95 .342
-.142 .408
Interaction
-.149 .209 -.71 .475
-.559 .261
PCL-R .571 .177 3.23 .001
.225 .918
Constant 1.754
.150 11.68
.000
1.460 2.048
Alpha .717 .264 - - .349 1.476LR test of alpha=0: chibar2(01) = 105.24 Prob >= chibar2 = 0.000
Comparison of model with and without PCL-R in it: Subtracting the R-
Square value for the model including psychopathy (Pseudo R square=.041)
as a predictor from the model that does not include it (Pseudo R
square=.007). So 0.041-0.007= 0.034. So the model with PCL-R in it
explains an additional 3.4% of the variance in convictions including
robbery.
145
This indicates that there was no support for the hypothesis that discrepant
self-esteem would predict higher levels of aggression controlling for
clinician-rated psychopathy. In fact, comparison of the two models indicates
that the model including clinician-rated psychopathy explains considerably
more variance in convictions including robbery than the model that does not
include psychopathy.
Incidents on the Ward
To explore whether fragile self-esteem would predict higher levels of
institutional aggression controlling for clinician-rated psychopathy, a
logistic regression was performed with group membership (no incidents in
the past 12 months or 1 or more incidents in the past 12 months) as the
binary dependent variable and explicit self-esteem (RSES scores), implicit
SE (IAT score), the interaction between these two variables (RSES*IAT)
and clinician-rated psychopathy (PCL-R scores) as continuous predictor
variables (see Table 17). The full model was statistically significant, (χ2(50,
4)=.15.558, p = .004), indicating that the model was able to distinguish
between those who had and had not been involved in an incident on the
ward. The model as a whole explained between 26.7% (Cox and Snell R
Square=.267) and 37.9% (Nagelkerke R square= .379) of the variance in
group membership, and correctly classified 80% of the cases, which is better
than the model’s predictive ability with only the constant in the equation
(70%). However, individually, only clinician-rated psychopathy made a
significant unique contribution (β=1.596, p<.01).
Table 17. Regression model predicting institutional aggression, controlling for psychopathy
146
Beta S.Error
Wald
p Exp(B)
95% CI interval for Exp(B)Lower
Upper
RSES -.628 .392 2.564
.109
.533 .247 1.151
IAT .315 .408 .598 .439
1.371 .616 3.050
Interaction
-.567 .472 1.442
.230
.567 .225 1.431
PCLR 1.596
.555 8.266
.004
4.931 1.662 14.632
Constant -1.224
.426 8.237
.004
.294 - -
Comparison of model with and without PCL-R in it: Subtracting the R-
Square value for the model including psychopathy (Cox and Snell=.267,
Nagelkerke= .379) as a predictor from the model that does not include it
(Cox & Snell=.049, Nagelkerke=.070). So 0.276-0.049=21.8% (Cox and
Snell) and 0.379-0.070=0.309 so the model with PCL-R in it explains an
additional 21.8% to 30.9% of the variance in institutional aggression
grouping.
This indicates that there was no support for the hypothesis that discrepant
self-esteem would predict higher levels of aggression controlling for
clinician-rated psychopathy. In fact, comparison of the two models indicates
that the model including clinician-rated psychopathy explains considerably
more variance in institutional aggression than the model that does not
include psychopathy.
Severity of Index Offence
To explore whether fragile self-esteem will predict higher severity of
aggression in the index offence, controlling for clinician-rated psychopathy
147
an ordinal regression analysis was performed to determine whether the odds
of committing a more aggressive index offence differed significantly
according to levels of explicit, implicit, discrepant (the explicit*implicit
interaction) self-esteem and clinician-rated psychopathy. Group membership
(index offence MOAS score lower than 16, MOAS score of 16 or MOAS
score of greater than 16) was the ordinal dependent variable and explicit
self-esteem (RSES scores), implicit self-esteem (IAT score), the interaction
(representing discrepant self-esteem) and clinician-rated psychopathy (PCL-
R scores) were the continuous predictor variables. The test of parallel lines
was not significant (χ2(4) = 4.312, p = .365) indicating that each
independent variable has an identical effect at each cumulative split of the
ordinal dependent variable, thus the assumption of proportional odds has
been met. The full model containing all predictor variables was not a
significantly better fit than the model without the predictor variables (χ2(4)
= 4.299, p = .367). The model as a whole only explained only 9.4%
(Nagelkerke Pseudo R square= .094) of the variance in group membership.
Findings indicate that for a one unit increase in explicit self-esteem, we
expect a .222 increase in the ordered log odds of being in a higher severity
of index offence group, however this finding was not significant (p=.427).
For a one unit increase in implicit self-esteem, we expect a .206 decrease in
the ordered log odds of being in a higher severity of index offence group,
however this finding was not significant (p=.460). For a one unit increase in
clinician-rated psychopathy, we expect a .027 increase in the ordered log
odds of being in a higher severity of index offence group, however this
finding was not significant (p=.924). For a one unit increase in
148
explicit*implicit self-esteem interaction, we expect a .622 increase in the
ordered log odds of being in a higher severity of index offence group,
however this finding was not significant (p=.113). This analysis indicates
that there was no support for the hypothesis that discrepant self-esteem will
predict higher severity of aggression in the index offence, controlling for
clinician-rated psychopathy.
Table 18. Regression model predicting severity of index offence, controlling for psychopathy
Estimate
S.Error
Wald p 95% CI
Lower
Upper
Threshold
MOAS Group 0
-1.457 .366 15.826
.000
-2.175
-.739
MOAS Group 1
.726 .310 5.475 .019
.118 1.334
Location
RSES .222 .280 .630 .427
-.327 .772
IAT -.206 .279 .545 .460
-.752 .340
PCLR .027 .283 .009 .924
-.527 .581
RSES* IAT Interaction
.622 .392 2.518 .113
-.146 1.390
Instrumental and Reactive Aggression
149
To test whether self-esteem fragility will predict whether the index offence
involved more instrumental or reactive aggression controlling for clinician-
rated psychopathy, a logistic regression was performed with group
membership (instrumental or reactive) as the binary dependent variable and
explicit self-esteem (RSES scores), implicit SE (IAT score), the interaction
between these two variables (RSES*IAT) and clinician-rated psychopathy
(PCL-R scores) as continuous predictor variables. The full model was not
statistically significant, (χ2(47, 4)=.2.425, p = .658), indicating that the
model was not able to distinguish between groups. Individually, none of the
predictor variables contributed significantly to the model (see Table 10).
Table 20. Regression model predicting instrumental or reactive aggression, controlling for psychopathy
Beta
S.Error
Wald
p Exp(B)
95% CI interval for Exp(B)Lower
Upper
RSES .106 .305 .120 .729
1.112 .611 2.021
IAT -.306
.326 .883 .347
.736 .389 1.394
Interaction
-.276
.400 .476 .490
.759 .347 1.661
PCLR .412 .335 1.517
.218
1.510 .784 2.909
Constant .430 .311 1.910
.167
1.537 - -
150
These findings do not support the hypothesis that self-esteem fragility
predicts whether the index offence involved more instrumental or reactive
aggression controlling for clinician-rated psychopathy.
Appendix 11
Regressions with physically aggressive convictions excluding robbery
Does self-esteem fragility predict higher levels of aggression, controlling
for psychopathy?
To explore whether fragile self-esteem would predict a higher number of
physically aggressive convictions (excluding robbery) controlling for
psychopathy, a zero-truncated negative binominal regression was used with
number of physically aggressive convictions excluding robbery as the
151
dependent variable and with explicit self-esteem (RSES scores), implicit SE
(IAT score), the interaction between these two variables and clinician-rated
psychopathy (PCL-R scores) as continuous predictor variables (see Table
21). The model as a whole was not statistically significant (LR χ2(4, N = 49)
= 5.44, p = .244), indicating that discrepant self-esteem did not predict a
higher number of physically aggressive convictions (excluding robbery)
controlling for psychopathy. The model as a whole explained only 2.6%
(Pseudo R2= .026) of the variance in counts of convictions. Only one
individual variable (PCL-R score) made a significant contribution to the
model (β =.305, p=.045). These results do not support the hypothesis that
discrepant self-esteem would predict a higher number of physically
aggressive convictions (excluding robbery) controlling for psychopathy,
because psychopathy was the only significant predictor. The alpha
dispersion parameter (alpha= .392) was significantly greater than zero
(likelihood ratio χ2 test that alpha equals 0 (df=1)= 12.61), which confirms
that the data are over dispersed and a negative binomial model provides a
better estimation of the data than a zero truncated poisson model.
Table 21. Regression model predicting number of physically aggressive convictions excluding robbery
Coef S.Error
z p 95% CI interval for Exp(B)
Lower
Upper
RSES .065 .140 .47 .641
-.210 .341
IAT .067 .126 .64 .591
-.179 .314
Interaction
-.193 .174 -1.11
.267
-.534 .148
PCL-R .305 .152 2.0 .04 .007 .602
152
1 5Constant 1.10
9.148 7.4
8.000
.818 1.399
Alpha .392 .227 - - .126 1.219LR test of alpha=0: chibar2(01) = 12.61 Prob >= chibar2 = 0.000
Does self-esteem fragility predict higher levels of aggression, controlling
for narcissism?
To explore whether fragile self-esteem would predict a higher number of
physically aggressive convictions (excluding robbery) controlling for
narcissism, a zero-truncated negative binominal regression was used with
number of physically aggressive convictions excluding robbery as the
dependent variable and with explicit self-esteem (RSES scores), implicit SE
(IAT score), the interaction between these two variables and narcissism
(NPI scores) as continuous predictor variables (see Table 22). The model as
a whole was not statistically significant (LR χ2(4, N = 47) =2.62, p =.622),
indicating that discrepant self-esteem did not predict a higher number of
physically aggressive convictions (excluding robbery) controlling for
narcissism. The model as a whole explained only .01% (Pseudo R2= .0131)
of the variance in counts of convictions. None of the individual variables
made a significant contribution to the model. These results do not support
the hypothesis that discrepant self-esteem would predict a higher number of
physically aggressive convictions (excluding robbery) controlling for
narcissism. The alpha dispersion parameter (alpha= .511) was significantly
greater than zero (likelihood ratio χ2 test that alpha equals 0 (df=1)= 16.32),
which confirms that the data are over dispersed and a negative binomial
model provides a better estimation of the data than a zero truncated poisson
model.
153
Table 22. Regression model predicting number of physically aggressive convictions excluding robbery, controlling for narcissism
Coef S.Error
z p 95% CI interval for Exp(B)Lower
Upper
RSES -.001 .166 -.01 .995
-.327 .325
IAT .099 .141 .70 .483
-.177 .374
Interaction
-.064 .189 -.34 .734
-.434 .306
NPI .189 .159 1.19
.236
-.123 .501
Constant 1.085
.170 6.39
.000
-.752 1.417
Alpha .511 .296 - - .164 1.591LR test of alpha=0: chibar2(01) = 16.32 Prob >= chibar2 = 0.000
To explore the Hypothesis 3 (psychopathy will predict higher levels of
aggression) in relation to physically aggressive convictions (excluding
robbery), a zero-truncated negative binominal regression was performed
with number of physically aggressive convictions excluding robbery as the
dependent variable and with PCL-R scores as the continuous predictor
variable (see Table 23). One outlier from the dependent variable was
removed, as discussed earlier.
The model as a whole was not statistically significant (LR χ2(1, N = 49) =
3.43, p = .064), indicating that psychopathy did not predict a higher number
of physically aggressive convictions (including robbery). The coefficient for
PCL-R was statistically significant suggesting that log count of convictions
increases by .274 for each unit increase PCL-R score (β=.274, p=.066), with
1.6% of the variance explained (Pseudo R2= .016). The alpha dispersion
parameter (alpha=-.453) was significantly greater than zero (likelihood ratio
χ2 test that alpha equals 0 (df=1)= 15.48), which confirms that the data are
154
over dispersed and a negative binomial model provides a better estimation
of the data than a zero truncated poisson model.
Table 23Zero truncated negative binomial regression predicting physically aggression convictions excluding robbery using psychopathy
Coefficient Standard Error
95% confidence intervals
P-value Pseudo R2
Likelihood ratio χ2
alpha = 0 (p)
PCL-R .274 .149 -1.18 to .566 .066 .016 15.48(p<0.001)Alpha .453 .255 .150 to 1.366
Does psychopathy predict higher counts of physically aggressive
convictions excluding robbery?
To explore Hypothesis 5 in relation to physically aggressive convictions
(excluding robbery), a zero-truncated negative binominal regression was
performed with number of physically aggressive convictions excluding
robbery as the dependent variable and with ESE ISE and the interaction
between these two variables as continuous predictor variables (see Table
24). The model as a whole was not statistically significant (LR χ2(3, N = 49)
= 1.36, p = .715), indicating that discrepant self-esteem did not predict a
higher number of log count of physically aggressive convictions (excluding
robbery). The model as a whole explained only .07% (Pseudo R2= .007) of
the variance in log counts of convictions. None of the individual variables
(including the interaction) made a significant contribution to the model. The
alpha dispersion parameter confirms that a negative binomial model
provides a better estimation of the data than a zero truncated poisson model.
Table 24
155
Zero Truncated Negative Binomial Regression Predicting Physically Aggression Convictions Excluding Robbery Using Explicit and Implicit Self-Esteem
Coefficient Standard Error
95% confidence intervals
P-value Pseudo R2
Likelihood ratio χ2
alpha = 0 (p)
RSES .073 .151 -.223 to .370 .627
.007 16.86(p<0.001)
IAT .109 .131 -.147 to .366 .403RSES*IAT -.094 .178 -.443 to .556 .599Alpha .501 .282 .116 to 1.512
Appendix 12
156
Study Protocol, Including Case Report Form (CRF) With all Details of Measures Used in the Study
The Relationship Between Self-Esteem Fragility and
Aggression in a High Security Hospital [version 3 14/09/15]
Sponsor: University of Surrey
IRAS project ID: 168429
REC Reference: 15/LO/0686
Funder: University of Surrey
Protocol version no: Version 3
Protocol version date: 14th September 2015
Contact Details
157
For general queries, supply of study documentation and central data management please contact:
Carly Samson
Trainee Clinical Psychologist
<name of hospital removed>
Tel: <removed>
Email: <removed>
To contact the lead principal investigator:
Dr Simon Draycott
Highly Specialist Clinical Psychologist
Address: <name of hospital removed>
Tel: <removed>
Email: <removed>
158
Research Team
1. Chief investigator Carly Samson, Trainee Clinical Psychologist, University of Surrey
2. Lead principal investigator Dr Simon Draycott and custodian of the data <name of hospital and address removed>
3. University supervisor Dr Simon DraycottSenior Tutor
<address removed>
Dr Erica Hepper
Lecturer & Senior Professional Training Tutor
<address removed>
159
Table of Contents
Frequently used abbreviations ……………………………………………………………………………… 6
1.0 Protocol summary
1.1 Overview …………………………………………………………………………………………… 7
1.2 Summary………………………………………………………………………………………….… 8
2.0 Introduction
2.1 Background…………………………………………………………………………………………. 9
2.2 Rationale …………………………………………………………………………………………… 10
3.0 Objectives
3.1 Primary aim …………………………………………………………………………………………
11
3.2 Secondary aim ……………………………………………………………………………………..
11
4.0 Methods
4.1 Study design and setting
…………………………………………………………………………. 11
4.2 Participants …………………………………………………………………………………………
11
4.3 Inclusion/exclusion criteria
……………………………………………………………………….. 12
4.4 Measures …………………………………………………………………………………………...
12
4.5 Procedure …………………………………………………………………………………………..
15
5.0 Informed consent procedure …………………………………………………………………………….
15
6.0 Statistics
6.1 Sample size calculation ………………………………………………………………………...
… 17
6.2 Data analysis…………………………………………………………………………………….….
17
7.0 Data management
7.1 Confidentiality ………………………………………………………………………………………
18
7.2 Data storage ………………………………………………………………………………………..
18
160
7.3 Data transfer ………………………………………………………………………………………..
19
7.4 Archiving …………………………………………………………………………………………… 19
8.0 Risks, burdens and benefits of participation
8.1 Risks and burdens for participants
…………………………………………………………….... 19
8.2 Benefits for participants
…………………………………………………………………………... 20
8.3 Risks for researchers ………………………………………………………………………...……
20
9.0 Withdrawal of participants ……………………………………………………………………………....
20
10.0 Assessment and management of risk ………………………………………………………. .20
11.0 Ethical and regulatory approvals
11.1Ethical approval ………………………………………………………………………..… 21
11.2Protocol amendments ………………………………………………………...………… 21
11.3Participant confidentiality and data protection
……………………………………….. 21
12.0 Sponsorship ……………………………………………………………………..………………. 21
13.0 Insurance …………………………………………………………………………………………. 22
14.0 Funding …………………………………………………………………………………………… 22
15.0 References ……………………………………………………………………………………….. 23
16.0 Appendices ………………………………………………………………………………………. 29
Frequently used abbreviationsPD Personality DisorderBPD Borderline Personality Disorder
161
NPD Narcissistic Personality DisorderCI Chief Investigator<name of trust removed>PI Principal InvestigatorRSES Rosenberg Self Esteem ScaleIAT Implicit Association TestNPI Narcissistic Personality InventorySRPS Self-Report Psychopathy ScaleANEW Affective Norms for English WordsMOAS Modified Overt Aggression ScaleCGVI Coding Guide for Violent IncidentsPCL-R Psychopathy Checklist-Revised
1.0 Protocol Summary
1.1 Overview
162
Full title The Relationship Between Self-Esteem Fragility and Aggression in Clinical Forensic Patients
Short title Self-Esteem and Aggression in Forensic Psychiatric Patients
Sponsor University of Surrey
Funder (reference) University of Surrey
REC <removed>
Design A questionnaire study using a cross-sectional correlation design
Primary aim To determine whether self-esteem fragility mediates the relationship between narcissism, psychopathy and aggression, which will enhance our understanding of risk factors associated with violent behaviour and potentially inform treatment approaches.
Secondary aims 1) To explore the relationship between self-esteem fragility and narcissism2) To explore the relationship between self-esteem fragility and psychopathy3) To determine whether the relationship between narcissism and aggression is explained by self-esteem fragility4) To determine whether the relationship between psychopathy and aggression is explained by self-esteem fragility
Inclusion criteria Male participants with a violent forensic history, who are fluent in English and sufficiently mentally stable to participate (this will be checked with the nurse in charge on the day)
Exclusion criteria Patients who are diagnosed with a developmental disorder or traumatic brain injury as this may affect their understanding of the questionnaires and computer task. Patients who have grandiose delusions (i.e. fixed beliefs) as this may have an impact on their sense of self.
Planned site 1 high security psychiatric hospital <name of hospital removed>
Recruitment period April – November 2015
1.2 Summary
163
Self-disturbances are prominent features of many mental health difficulties (Edwards et al., 2012; Fonagy, 1999; Bateman & Fonagy, 2004), and have been highlighted as important risk factors in violent and aggressive behaviour (Baumeister et al., 1996; Donnellan et al., 2005; Kantor & Jasinski, 1998). Self-esteem has consistently been linked to aggression, and research within the general population indicates that fragile self-esteem in particular (a discrepancy between unconscious and conscious evaluations of self-worth) is an important mediator of aggressive behaviour (Kernis, 2003; Kernis, 2005; Edwards and Bond, 2012). Dissonance between conscious and unconscious evaluations of the self have been identified in people with high levels of narcissistic and psychopathic traits (Campbell & Foster, 2007; Zeigler-Hill, 2006a; Cleckley, 1941), and at a clinical level (seen in personality disorders including Narcissistic Personality Disorder and Psychopathy) these traits are linked to violence and aggressive behaviour. Indeed, there is a very high prevalence of people with PD diagnoses in forensic populations (Fazel & Danesh, 2002; New et al., 2004). Therefore it is possible that clinical levels of self-disturbance are related to an increased risk of violence behaviour. The purpose of this is to explore whether self-esteem fragility mediates the relationship between narcissism and psychopathy and aggression in a psychiatric sample of people with violent forensic histories, who will be recruited from <name of hospital removed> a high security forensic hospital in <location removed>. Participants will be asked to complete some self-report questionnaires to measure levels of explicit self-esteem, narcissistic and psychopathic traits, and to complete a short computerised test to measure levels of implicit self-esteem. Information pertaining to violence forensic history and institutional aggression will be obtained from participants’ medical files. Findings could be used to predict future risk of violence and reoffending and will have important implications for the treatment of offenders.
2.0 Introduction
164
2.1 Background
Three key aspects of self appear to be related to aggressive behaviour in samples of people with personality disorder (PD) diagnoses and/or forensic histories: self-esteem, self-concept and identity diffusion. The existing evidence base appears to be compromised by poorly defined aspects of self, making it difficult for one to identify exactly what is being measured, as well as one-dimensional measures of rather complex, multi-dimensional constructs. Therefore studies employing more rigorous, in-depth measures of aspects of self are urgently required.
The role of the self in social and psychological functioning is well documented (Campbell, Assanand & Di Paula, 2000). Self-disturbances are prominent features of many mental health difficulties (Edwards et al., 2012; Fonagy, 1999; Bateman & Fonagy, 2004), and have been highlighted as important risk factors in violent and aggressive behaviour (Baumeister et al., 1996; Donnellan et al., 2005; Kantor & Jasinski, 1998). In particular, self-esteem is an important aspect of self that has been linked to aggression, although whether it is high or low self-esteem is a contentious issue with research finding that both low and high self-esteem are related to violence (Baumeister, Smart & Boden, 1996; Baumeister, Bushman & Campbell, 2000; Boden, Fergusson, & Horwood, 2007). These conflicting findings may be explained by the consideration of self-esteem as a unilateral construct, using only explicit, self-report measures, which do not account for the unconscious, implicit component of self-esteem. Dual processing models demonstrate that information can be processed implicitly and explicitly simultaneously, with the former being outside of conscious awareness (Greenwald & Banaji, 1995). According to The Full Discrepancy Model (Figure 1), dissonance between implicit and explicit self-esteem generates a tenuous evaluation of self-worth, and it is this discrepancy that is an important mediator of aggressive behaviour (Kernis, 2003; Kernis, 2005; Edwards and Bond, 2012).
The Partial-Discrepancy model emphasises the low implicit/high explicit self-esteem cell in people with high levels of narcissistic traits and suggests that this particular discrepancy is linked to higher levels of aggressive behaviour. An explanation for this relationship is that people with discrepant self-esteem characterised by high explicit and low implicit attitudes (fragile self-esteem) towards the self lack a secure, intrinsic base for their positive self-evaluations. Therefore they react strongly to negative feedback and use maladaptive defences such as aggression to attack the perceived source of the threat in order to defend their positive, but vulnerable, feelings of self-worth (Kernis, Granneman et al., 1989). This pattern of explicit, grandiose evaluation of the self concealing unconscious negative feelings of self-worth are consistent with current models of narcissism (Campbell & Foster, 2007; Zeigler-Hill, 2006a) and psychopathy (Cleckley, 1941). Little is known about the impact of high implicit, low explicit (damaged) self-esteem of aggressive behaviour, and research is needed to explore the direction of the discrepancy between implicit and explicit self-esteem.
Most of the research into the relationship between fragile self-esteem and aggression come from studies of non-clinical and non-forensic populations. Despite the developments in our understanding of the importance of implicit, as well as explicit processes involved in self-esteem, research in clinical and forensic psychiatric settings are still using uni-dimensional, self-report measures of self-esteem. This is surprising as self-esteem fragility is a central component of narcissism and psychopathy (APA, 2013; Cleckley, 1941; Hare, 2003), both of which are consistently associated with aggression in research (for reviews see Reidy, Shelley-Tremblay & Lilienfeld, 2011; Roberts & Coid, 2007 & Baumeister et al, 2000). Therefore research is needed to explore whether self-esteem fragility mediates the relationship between narcissism and psychopathy and aggression in forensic psychiatric samples.
There is a high prevalence of people with diagnoses that are characterised by an enduring, fragile sense of self-worth (such as Narcissistic PD and Psychopathy) in forensic populations (Fazel & Danesh, 2002; New et al., 2004), so it is possible that
165
clinical levels of self-disturbance are related to an increased risk of violence behaviour. Psychopathy and narcissism share a number of traits including a lack of empathy, grandiosity, and exploitation of others, with some researchers describing psychopathy as an extreme form of pathological narcissism, and a type of narcissism particularly related to violence (Kernberg, 1975; Cale & Lilienfeld, 2006). It is therefore possible that self-esteem fragility mediates the relationship between narcissism, psychopathy and aggression. There are longstanding debates regarding the validity and reliability of categorical PD diagnoses in forensic populations (Sarkar & Duggan, 2010), therefore a more suitable approach may be to examine underlying traits of personality factors linked to aggression.
Findings could be used to predict recidivism and future risk of violence therefore research is urgently needed in this area. These findings also have important implications for the treatment of offenders, suggesting that, rather than delivering treatment programs that target the level of offenders’ self-esteem, treatment plans should be developed aiming to enhance congruence between conscious and unconscious feelings of self-worth.
2.2 Rationale for the present study
If self-esteem fragility underlies aggressive behaviour, then it may be hypothesised that aggression will increase as implicit self-esteem decreases, but only when explicit self-esteem is high (representing self-esteem fragility). Therefore, if it were possible to determine the degree of fragility of a patient’s self-esteem, it may be possible to identify those who are most likely to re-offend and offer further/alternative treatment targeted to improve one’s overall sense of self-worth. The literature suggests a strong link between narcissism and psychopathy and self-esteem fragility. Currently, the mechanism(s) by which this effect mediates violent behaviour is not fully understood, so exploring how and when self-esteem fragility mediates the effect of narcissism on aggression, and the effect on psychopathy on aggression, may result in a deeper understanding of this phenomenon, and may give insight into how that understanding can be applied to managing risk and treating violent offenders with PDs.
Due to the unreliability and difficulty with diagnosing these PDs in forensic populations (Sarkar & Duggan, 2010), measures of underlying traits of psychopathy and narcissism will be more appropriate for this study. They have the additional benefit that they require less participant and clinician time to administer compared to lengthy diagnostic interviews.
2.2.1 Hypotheses
1. Aggression (higher severity of index offence violence, greater of violent crimes committed, higher number and greater severity of aggressive incidents on the ward in the past 12 months) will increase as implicit self-esteem decreases, but only when explicit self-esteem is high (this will represent self-esteem fragility)
2. Self-esteem fragility will predict higher levels of narcissism and psychopathy
3. Self-esteem fragility will mediate the effect of narcissism on aggression4. Self-esteem fragility will mediate the effect of psychopathy on aggression
3.0 Objectives
3.1 Primary aim
To determine whether self-esteem fragility mediates the relationship between narcissism, psychopathy and aggression, which will enhance our understanding of risk factors associated with violent behaviour.
Secondary aims
166
1) To explore the relationship between self-esteem fragility and narcissism
2) To explore the relationship between self-esteem fragility and psychopathy
3) To determine whether the relationship between narcissism and aggression is explained by self-esteem fragility
4) To determine whether the relationship between psychopathy and aggression is explained by self- esteem fragility
4.0 Methods
4.1 Study design and setting
This research will take place at <name of hospital removed> which is a high secure psychiatric hospital. Relationships between personality constructs (narcissism, psychopathy, implicit and explicit self-esteem) and past history of aggression will be explored using correlational and regression analyses to test all of the experimental hypotheses (above) using a cross-sectional correlational design.
4.2 Participants
Screening of potential participants will be carried out by the PI, who is part of the existing care team at <name of hospital removed>, or the clinical team, to identify patients who are eligible to participate, and will determine whether they wish to be approached for research. The CI will meet with patients who wish to be approached for research to explain the study and give them an information sheet. Data from <name of hospital removed> indicates that there are approximately 150 people who may meet the inclusion criteria for this study. A recent study conducted with a similar sample investigating a similar topic (Edwards & Bond, 2012) reported that 55% of eligible participants agreed to participate. If 55% of the available population decided to take part that would yield a potential sample of 83 participants. Therefore, a sufficient number of participants are likely to take part to meet minimum sample size recommended by the power calculation (55 participants). Recruitment will be stopped once 60 participants have taken part, otherwise recruitment will be stopped at the end of the data collection period (30th November 2015).
4.3 Inclusion/exclusion criteria
Male participants with a violent forensic history, who are fluent in English and sufficiently mentally stable to participate (this will be checked with the nurse in charge on the day) will be included in the study.
Patients who are diagnosed with a developmental disorder or traumatic brain injury will be excluded as these difficulties may affect their understanding of the questionnaires and computer task. Patients who have grandiose delusions (i.e. fixed beliefs) will be excluded because these fixed beliefs have been found to reflect specific patterns of self-esteem (for a review see Kesting & Lincoln, 2013).
4.4 Measures
4.4.1 Implicit Self-Esteem
The Implicit Association Test (IAT, Greenwald, McGhee & Schwartz, 1998) will be used to measure implicit self-esteem. It is a computerised task that
167
measures unconscious associations between self-relevant and non-self-relevant words with pleasant and unpleasant words. The task will be run on an NHS laptop meeting the necessary security requirements of <name of hospital removed>
The IAT has been chosen as the measure of implicit self-esteem because it is the most widely used measure and it has shown the highest reliability of all the available measures of implicit self-esteem (Bosson et al, 2000; Krause et al. 2011). Scoring will be based on the IAT d calculation described in Greenwald, Nosek and Banaji (2003). Negative values of d signify low implicit self-esteem whereas high values indicate high implicit self-esteem.
The words used by Gregg & Sedikides in the original self-esteem IAT measure (2010) may not be appropriate for participants at <name of hospital removed> because they are likely to strongly associate “self” words to the word “murder” due to their forensic history rather than because of low-implicit self-esteem. Feedback from patients at <name of hospital removed> highlighted that some patients experiencing paranoia and/or from the Muslim community may be distressed by the word “bomb”. Therefore the words “bomb” and “murder” from the unpleasant word category have been replaced with two other words of equal valence, arousal and dominance using The Affective Norms for English Words (ANEW; Bradley & Lang, 2010).
i. Explicit Self-EsteemThe Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) consists of 10 self-report questions measuring global self-esteem with a maximum score of 30. A higher score indicates higher self-esteem. The scale has been validated for use with a variety of clinical groups, with satisfactory reliability and validity reported in a prison sample (Cale & Lilienfeld, 2006). This measure has been chosen as it is the most widely used measure of explicit self-esteem and would allow me to compare my findings with those from other studies in this area. A copy of the measure can be found in Appendix 2.
4.4.3.Fragile Self-Esteem
Fragile self-esteem (the discrepancy between implicit and explicit self-esteem), as well as the direction of the discrepancy in relation to aggression will be calculated as follows:
1. The main effects for implicit (IAT scores) and explicit self-esteem (RSES scores) on violent history ratings (i.e. severity of index offence and frequency and severity of previous violent convictions scored using the MOAS) and institutional violence ratings (number of violent incidents on the ward in the past 12 months and average severity scored using the MOAS) will be investigated using regression analyses.
2. Then the interaction of implicit and explicit self-esteem will be entered into the regression model.
3. Follow up simple slopes tests (Aiken & West, 1991) will be performed to explore the direction of the explicit and implicit self-esteem interaction.
This is the most common method used in studies exploring the fragility of self-esteem (Jordan, Spencer, & Zanna, 2005; Jordan, Spencer, Zanna, Hoshino-Browne, & Correll, 2003; Kernis et al., 2005).
4.4.4 Psychopathy
Underlying traits of psychopathy and narcissism will be measured rather than selecting a sample of people with related diagnoses because of the unreliability and difficulty with diagnosing these disorders in forensic
168
populations (Sarkar & Duggan, 2010). The Levenson Self-Report Psychopathy Scale (SRPS; Levenson et al, 1995) is a 26 item self-report measure which is designed to assess personality and behavioural traits associated with psychopathy. Participants are required to rate the extent to which they agree or disagree with statements using a four point scale. Higher scores are associated with higher levels of psychopathy. This measure constitutes two factors; Primary Psychopathy which measures a manipulative, callous interpersonal style and Secondary Psychopathy which assesses behavioural aspects of psychopathy such as inability to learn from mistakes and impulsivity. These two factors correspond approximately to the two factors of the Hare Psychopathy Checklist (PCL-R; Harpur, Hare & Hakistan, 1989). The SRPS was chosen to measure psychopathy because it is the least time consuming for participants compared to other lengthy scales, yet retains good convergent and discriminative validity (Sellbom, 2011). Levenson et al. (1995) found good internal consistency for the scale.
Individuals receiving support from forensic psychiatric services, such as patients at <name of hospital removed>, are likely to distort their responses on self-report rating scales (Whyte, Fox & Coxell, 2006). Indeed, some participants have stated outright that they are not responding honestly on the self-report measure. Therefore a more reliable assessment of psychopathy will be used in addition to the SRPS. The Psychopathy Checklist (PCL; Hare, 2003) is one of the most widely used measures of psychopathy with substantial supporting research (Hare, 2003). PCL-R ratings are available in the notes of most patients at <name of hospital removed>. For patients where this is not available, ratings will be made by file review by Dr Simon Draycott, the Principal Investigator for this project. This will allow the collection of levels of psychopathic traits that are less likely to be biased by socially desirable responding. <name of hospital removed> have a licence to use the PCL-R for research and clinical purposes.4.4.5 Narcissism
The Narcissistic Personality Inventory (NPI) is a 40 item self-report scale designed to assess narcissistic traits (Raskin & Hall, 1979). This scale is widely used in social science and personality research and has good internal consistency (Raskin & Terry, 1988). NPI scores correlate significantly with observer and self-report measures of narcissism (Raskin & Terry, 1988), indicating satisfactory validity.
4.4.6 Outcome Measures: Aggression
All measures will relate to direct, physical aggression only.
4.4.7 Frequency Counts
The number of previous physically aggressive convictions, and the number of physically violent incidents on the ward over the past 12 months will be counted. This information will be obtained from clients’ medical files.
4.4.8 Severity of Aggressive History
The Modified Overt Aggression Scale (MOAS; Kay, Wolkenfeld & Murrill, 1988) assesses four types of aggressive behaviour; verbal aggression, aggression against property, auto-aggression and physical aggression over the past week. Items are scored on a five point scale with high scores indicating more aggression (see Appendix 6). This scale will be used to rate the severity of violence of the index offence. Kay et al. (1988) reported good inter-rater reliability (r=0.94), and good discriminative validity for distinguishing high, intermediate and low violent groups. Twenty percent each of scored MOAS forms for index offence will be randomly selected and passed to the PI to rate in order to establish the degree of inter-rater reliability.
169
4.4.9 Severity of Institutional Aggression
The degree of violence of incidents on the ward over the past 12 months will be rated using the MOAS (described above). A MOAS form will be completed for each incident. Twenty percent each of scored MOAS forms for incidents on the ward will be randomly selected and passed to the PI to rate to establish the degree of inter-rater reliability.
Each participant will complete the IAT first to minimise the risk that other measures will prime their responses on the IAT. The sequence of completion of other measures (RSES, NPI and SRPS) will be counterbalanced using the Latin Square technique to reduce the risk of order effects.
4.4.10 Degree of Instrumental and Reactive Aggression
In order to increase understanding of how self-esteem might relate to different types of aggressive behaviour, Cornell et al’s (1996) Coding Guide for Violent Incidents (CGVI) will be used to rate the extent to which participants’ offences involved instrumental or reactive aggression.
This measure is used to code the presence of instrumental and reactive aggression on a four point scale. This measure has been used in forensic psychiatric settings (Laurell, Belfrage & Hellstrom, 2010) and has satisfactory inter-rater reliability (kappa coefficient of .85; Cornell et al. 1996). This will enable the researchers to determine how self-esteem may be linked to different types of aggression. Participants have already given consent for information from their medical files to be used for the purposes of this research project (see consent form v3 01.05.15).4.5 Procedure
The clinical team and site-specific PI, who is part of the clinical team, will identify patients who are eligible to participate in the present study and determine whether they wish to be approached for research purposes. Patients who consent to be approached for research will be provided with an information sheet and given the opportunity to ask questions with the CI or PI. Patients will be given at least 24 hours to decide whether they wish to participate. Patients who decide to participate will be invited to meet the CI who will obtain informed consent and then administer the computerised IAT, followed by the counterbalanced questionnaires. Participants will meet with the CI on one occasion lasting up to one hour in a quiet room adjacent to their ward. By signing relevant statement on the consent form, participants will be giving the CI permission to access their medical files in order to obtain details of their violent forensic history, as well as details of violent incidents on the ward, which will be counted and then scored by the CI using the MOAS and the CGVI. Twenty percent of MOAS forms and CGVI forms for violent convictions and for violent incidents on the wards will be randomly selected and rated by the PI to establish inter-rater reliability. PCL-R ratings from participants’ medical files will also be obtained, and if they are not available the PI (Dr Simon Draycott) will complete a PCL-R for research purposes by file review. Once all data has been collected and analysed, participants will be given a summary sheet explaining the overall findings of the study.
5.0 Informed consent procedure
Patients who have been identified by the clinical team at <name of hospital removed>as being eligible to participate in the present study, who have given consent to be approached for research, and who are considered to have capacity to consent by their responsible clinician (RC) or the nurse in charge, will be given an information sheet to read and consider whether they would like to take part. The information sheet provides information
170
about the study and what participation would involve. The information sheets have been designed and developed to make them as useful and accessible as possible. The CI has collaborated with service users at the <name of hospital removed> Patient Forum for feedback regarding the clarity and acceptability of the information sheets.
Patients will be given the opportunity to ask questions if they would like more information or if anything is unclear. They will be reminded that participation is entirely voluntary and their care will not be affected if they decide not to take part. Participants can consult with others if they wish. After at least 24 hours, patients will be asked for consent to participate in the prospective study by the CI. A refusal to participate will not affect their care or support in any way.
If patients decide that they wish to participate, they will be asked to sign a consent form, copies of which will be included in their medical file, in a locked research paperwork cabinet at <name of hospital removed> and given to participants if they wish. Written informed consent must be obtained before any study specific procedures are conducted.
Participants will be reminded that they can withdraw their consent at any time during data collection, and that all information obtained will be kept confidential and anonymous, unless they reveal or disclose information that may indicate a risk of harm to themselves or others, or a previous crime of which they have not been convicted. If this happens then the CI will inform the patient’s clinical team who will follow it up. This confidentiality caveat is stated in the participant information sheet.
In seeking consent, the research team will ensure that patients are making their decision on an informed basis. There are some misunderstandings that could lead to participants giving consent on an uninformed basis, which the research team need to be aware of:
1. Patients may believe that they are obliged to participate because the study is being conducted at <name of hospital removed> and thus may not understand that participation is voluntary.
2. They may think that if they refuse to take part or if they withdraw consent, there will be negative consequences for their treatment.
To guard against these misunderstandings, the CI will go through the information with the patient to ensure they understand what is involved in the study and the research team will adhere to the following guidelines:
1. The decision of the participant to withdraw consent or refuse to participate should be respected at all times
2. If the participant consistently refuses to make an appointment with the CI, or does not attend three consecutive appointments, this should be interpreted as their not wanting to be involved in the study.
The CI will be responsible for:
Checking that the information on the consent form is complete and legible
Checking that the participant has completed all relevant sections and signed and dated the form
Ensuring that the original signed consent form is stored at <name of hospital removed> and that a copy is given to the participant if requested
6.0 Statistics
171
6.1 Sample size calculation
Previous studies looking at the effect of self-esteem on aggressive behaviour have presented effect sizes of r=-0.31 in a forensic sample (Fruehwald et al. 1996) and r=-0.36 in an undergraduate sample (Falkenbach et al, 2013). Assuming a power of 0.8 to detect an effect size of 0.335 (average of the two effect sizes), one-tailed with alpha=0.05 using a correlational test, an a priori calculation using G*Power 3.1.7 (Faul, Erdfelder, Lang & Buchner, 2007) suggested a sample size of 51 is needed.
Data from <name of hospital removed> indicates that there are 150 people who may meet the inclusion criteria for this study (personal communication from the PI). A recent study conducted with a similar sample investigating a similar topic (Edwards & Bond, 2012) reported that 55% of eligible participants agreed to participate. So if 55% of the available population decided to take part that would yield a sample of 83 participants. Therefore, a sufficient number of participants are likely to take part to meet minimum sample size recommended by the power calculation. The CI will aim to recruit 60 participants.
6.2 Data analysis
In the first instance, correlational analyses will be performed to explore relationships between all variables. Moderation and mediation analyses will then be performed to test the hypotheses as outlined below:
6.2.1 Moderation analyses
1. The main effects for implicit (IAT scores) and explicit self-esteem (RSES scores) on violent history ratings (i.e. severity of index offence and frequency and severity of previous violent convictions scored using the MOAS), institutional violence ratings (number of violent incidents on the ward in the past 12 months and average severity scored using the MOAS) and ratings of instrumental/reactive aggression will be investigated using regression analyses.
2. Then the interaction of implicit and explicit self-esteem will be entered into the regression model.
3. Follow up simple slopes tests (Aiken & West, 1991) will be performed to explore the direction of the explicit and implicit self-esteem interaction.
This will allow the investigation of Hypothesis One; that increased aggression will increase as self-esteem decreases, but only when explicit self-esteem is high (representing self-esteem fragility). This is the most common method used in studies exploring the fragility of self-esteem (Jordan, Spencer, & Zanna, 2005; Jordan, Spencer, Zanna, Hoshino-Browne, & Correll, 2003; Kernis et al., 2005).
The analyses above will be repeated to investigate the extent to which the interaction between explicit and implicit self-esteem predict scores on the PCL-R and NPI. This will allow Hypothesis Two to be tested; that self-esteem fragility will predict higher levels of narcissism and psychopathy.
6.2.2 Mediation analyses
A freely available computational tool for SPSS; The Hayes PROCESS method (Hayes, 2012), will be used to explore the mediating effect of self-esteem fragility on narcissism to predict aggression. This would allow the testing of Hypothesis Three; that esteem fragility will mediate the effect of narcissism on aggression.
172
The same method will be used to explore the mediating effect of self-esteem fragility on psychopathy to predict aggression, allowing the testing of Hypothesis Four, that self-esteem fragility will mediate the effect of psychopathy on aggression.
7.0 Data management
7.1 Confidentiality
All data will be handled in accordance with the Data Protection Act (1998). Information obtained will be kept confidential, unless participants reveal or disclose information that may indicate a risk of harm to themselves or others, or a previous crime of which they have not been convicted. If this happens then the CI will inform the patient’s clinical team who will follow it up. This confidentiality caveat is stated in the participant information sheet.
The questionnaires completed by participants for the study will not bear the participant’s name or other personal identifiable data. Participants’ electronic patient record system (RiO) number and unique identification number will be used for identification in order to match their data to details of violent convictions and aggressive incidents. Details of forensic history and aggressive behaviour on the ward will be coded using the MOAS, so participants cannot be identified from their criminal history.
Personal data will only be accessed by the direct clinical care team at <name of hospital removed>, the CI and PI. This data used for the research study will be stored in locked cabinets at <name of hospital removed>
Only anonymised data will be stored on password protected computers and encrypted password-protected USB sticks.
7.2 Data storage
Names of eligible participants will only be passed from the clinical team and PI, who have access to participants’ personal data as part of routine clinical practice, to the CI once consent has been given by patients to be approached for research. Names and wards where patients are receiving care will be requested from participants for the CI to make contact with participants, obtain informed consent and collect data. Hard copies of consent forms and questionnaires will be kept in a locked cabinet at <name of hospital removed>. The key will be kept separately from the cabinet in a locked office at <name of hospital removed>. Data kept electronically will be anonymised by the same identification number, and stored securely on a password-protected NHS computer at <name of hospital removed>. Details of violent criminal history and aggression on the wards will be coded using the MOAS, so participants cannot be identified from their criminal history. Anonymised data will be transferred using an encrypted, password protected USB stick for data analysis purposes to a computer which only the research team have access and will only be stored for the time for which it is required. The requirements of the Data Protection Act (1998) will be complied with throughout the research. No personal data will be shared with any other person or organisation unless participants reveal or disclose information that may indicate a risk of harm to themselves or others, or a previous crime of which they have not been convicted. If this happens then the CI will inform the patient’s clinical team, as would happen in any meeting with a staff member, who will follow it up, as stated in the participant information sheet. Once the project has finished, anonymised data will be stored for at least 10 years using a secure data archiving system in line with the University of Surrey’s data retention and storage policies.
7.3 Data transfer
173
Anonymised data will be transferred using an encrypted, password protected USB stick for data analysis purposes to a computer which only the research team have access and will only be stored for the time for which it is required.
E-mail will routinely be used among the research team to correspond about the study. However, names and details of participants will never be used or sent in e-mail correspondence, instead unique identification numbers will be used if necessary. No data collected in connection with the study will be sent by e-mail at any time, unless fully anonymised. Moreover, e-mail accounts at sites and at the University of Surrey will be accessible only with a personal password.
7.4 Archiving
In accordance with its current Records Retention Schedule, research data are retained at the University of Surrey for at least 10 years after the research has ended. Data will be treated as described in ethics regulations and consent agreements signed by the participants. Anonymised data will be stored using one of the secure offsite data repositories recommended by the University of Surrey in the University’s Research Data Management Policy where access to stored records is strictly controlled.
8.0 Risks, burdens and benefits of participation
8.1 Risks and burdens for participants
There is a risk that the negative words displayed during the implicit association test may distress participants. Advice has been sought from patients at <name of hospital removed> regarding the likelihood of distress and whether any adaptations need to be made. Two words were amended as a result of discussions with patients as outlined in Section 4.4.1. The content of the questionnaires will be explained to participants before they complete them and they will be reminded of their right to withdraw at any time so that participants who believe they will find such topics difficult need not complete them. Every effort will be made to put participants at ease. If participants do become distressed during the study support will be available at all times through their care team, who will be located nearby at all times on the wards, who can talk to them and deal with any issues that may arise as per <name of hospital removed> policies and procedures. The CI has attended a <name of hospital removed> induction to ensure they are familiar with all policies and protocols for managing such situations.
As part of the study, participants will be providing personal data and will agree for the CI to have access to their medical records. There is a potential risk of breakdown of confidentiality, although this risk is low as stringent data handling safeguards will be followed to ensure confidentiality. The risk will be minimised by following stringent data handling safeguards. Data will be stored in accordance with the Data Protection Act (1998) and all patient identifiable information will be kept in locked filing cabinets or on NHS computers at <name of hospital removed>, coded using their unique participant identification number and electronic patient record system number. All outcome data taken offsite for analysis will be anonymised, password protected and kept on encrypted devices or NHS/University of Surrey secure computer networks.
Deciding whether or not to participate in this study may pose an opportunity cost to participants if they are undertaking paid work at <name of hospital removed>. Therefore, participants will be reimbursed an amount approximately equivalent to what they would be paid for a half day of work at the hospital to compensate them.
174
8.2 Benefits for participants
There are no anticipated benefits to research participants, although each participant will be compensated £3 each for their time. This is to reimburse them for their time and to compensate them an equivalent amount for missing out on the opportunity for paid work. The money will be paid electronically into participants’ hospital accounts, as patients are not allowed to have cash at <name of hospital removed>
8.3 Risks for researchers
<name of hospital removed>is a high secure psychiatric hospital supporting people who pose a high risk to themselves and to others. In order to manage the risk to the researchers, the CI will attend a full staff induction at <name of hospital removed>including Breakaway training. As an existing member of staff at <name of hospital removed>, the PI attends Breakaway and refresher training yearly. The CI, who will be in charge of data collection, will be accompanied by other member of staff, or observed by permanent members of staff when with patients. There are alarms in every room at the hospital, and the CI will ensure she is positioned within reaching distance of an alarm when with patients.
9.0 Withdrawal of participants
If a participant wishes to withdraw from the study during the data collection, their decision will be respected and no further data will be collected. If a participant wishes to withdraw after data has been collected, then then all data collected up to their withdrawal will still be used, as explained in the participant information sheet.
10.0 Assessment and management of risk
The CI and PI are employed by NHS services and are therefore already familiar with the procedures of seeing patients with aggressive histories, and in assessing risk. The CI has attended a full staff induction for <name of hospital removed>, including Breakaway training and <name of hospital removed> policies and procedures around managing risk, before commencing data collection. As an existing member of staff at <name of hospital removed>, the PI attends Breakaway and refresher training yearly. The CI, who will be in charge of data collection, will be accompanied by other member of staff, or observed by permanent members of staff when with patients. There are alarms in every room at the hospital, and the CI will ensure she is positioned within reaching distance of an alarm when with patients. The CI’s academic supervisor will have no contact with participants at <name of hospital removed>
11.0 Ethical and regulatory approvals
In conducting this study the Sponsor, University of Surrey and <name of hospital removed> shall comply with all laws and statutes, as amended from time to time, applicable to the performance of research studies with human participants including, but not limited to:
The Human Rights Act (1998) The Data Protection Act (1998) The Research Governance Framework for Health and Social Care,
issued by the UK Department of Health (Second Edition, 2005).
11.1 Ethical Approval
175
The study will be conducted in accordance with the terms and conditions of the ethical approval given to the trail. The study is seeking ethical approval from <name of ethics committee removed>
11.2 Protocol amendments
The University of Surrey will be responsible for gaining ethical approval for all amendments made to the protocol and other trial-related documents. Once approved, the University of Surrey will ensure that all amended documents are distributed to <name of hospital removed>
11.3 Participant confidentiality and data protection
The University of Surrey will preserve participant confidentiality and will not disclose or reproduce any information by which participants could be identified. Data will be stored in a secure manner in accordance with the Data Protection Act (1998).
12.0 Sponsorship
12.1 Sponsor details
Sponsor name: University of Surrey
Address: <address removed>
Contact <name removed> (Research Integrity & Governance Officer). <contact details removed>
13.0 Insurance
A copy of the insurance certificate has been submitted with the REC application.
14.0 Funding
<name of trust removed> NHS Trust are funding the CI in their capacity as a trainee clinical psychologist, this research study is part of that training.
15.0 References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Newbury Park, CA: Sage Publications.
American Psychiatric Association, (2013). Diagnostic and Statistical
Manual of Mental Disorders (5th ed). Arlington, VA: American
Psychiatric Publishing.
Bateman, A.W., & Fonagy, P. (2004). Mentalization-based treatment of
BPD, Journal of Personality Disorders, 18, 36-51
Baumeister, R. F., Bushman, B. J., & Campbell, W. (2000). Self-esteem,
narcissism, and aggression: Does violence result from low self-
176
esteem or from threatened egotism? Current Directions in
Psychological Science 9(1), 26-29.
Baumeister, R. F., Smart, L. & Boden, J.M. (1996). Relation of threatened
egotism to violence and aggression: The dark side of high self-
esteem, Psychological Review 103(1), 5-33.
Baumeister, R.F., Smart, L., & Boden, J.M. (1996). Relation of threatened
egotism to violence and aggression: The dark side of high self-
esteem, Psychological Review, 103(1), 5-33.
Bentall RP, Kinderman P, Kaney S (1994) The self, attributional processes
and abnormal beliefs: Towards a model of persecutory delusions,
Behaviour, Research & Therapy, 32, 331–341.
Blackburn, R & Glasgow, D. (2006). Manual for the Chart of Interpersonal
Reactions in Closed Living Environments (CIRCLE). University of
Liverpool.
Boden, J. M., Fergusson, D.M., & Horwood, L. (2007). Self-esteem and
violence: Testing links between adolescent self-esteem and later
hostility and violent behaviour, Social Psychiatry and Psychiatric
Epidemiology 42(11), 881-891.
Bradley, M.M. & Lang, P.J. (2010). Affective Norms for English Words
(ANEW): Instruction manual and affective ratings. Technical
Report C-2. University of Florida, Gainesville, FL.
Buss, A.H., & Perry, M. (1992). The aggression questionnaire, Journal of
Personality & Social Psychology, 63(3), 452-459.
Cale, E. M. & Lilienfeld, S.O. (2006). Psychopathy factors and risk for
aggressive behavior: A test of the "threatened egotism"
hypothesis, Law and Human Behavior 30(1), 51-74.
177
Campbell, J.D., Assand, S., & Di Paula, A. (2000). Structural features of
the self-concept and adjustment, In A. Tesser, R.B. Felson, & J.M.
Suls, (Eds). Psychological Perspectives on Self and Identity (pp
115-138). Washington, DC, US: American Psychological
Association.
Campbell, K.W., & Foster, J.D. (2007). The narcissistic self: Background,
an extended agency model, and ongoing controversies. In C.
Sedikides & S.J. Spencer (Eds) Frontiers in Social Psychology:
The Self (pp 115-138). UK: Psychology Press.
Cleckley, H. (1941) The So-called Psychopathic Personality, With Special
Emphasis on his Status in the Selective Service, Journal of the
Medical Association of Georgia, 30, 466–472.
Cornell, D. G., Warren, J., Hawk, G., Stafford, E., Oram, G., & Pine, D.
(1996). Psychopathy of instrumental and reactive violent offenders.
Journal of Consulting and Clinical Psychology, 64, 783-790.
Dolan, M., & Blackburn, R. (2006). Interpersonal factors as predictors of
disciplinary infractions in incarcerated personality disordered
offenders, Personality and Individual Differences, 40, 897–907
Donnellan, M.B., Trzesniewski, K.H., Robins, R.W., Moffitt, T.E., & Caspi,
A. (2005). Low Self-Esteem Is Related to Aggression, Antisocial
Behavior, and Delinquency, Psychological Science, 16(4) ,328-
335.
Edwards, R. & A. J. Bond (2012). Narcissism, self-concept clarity and
aggressive cognitive bias amongst mentally disordered offenders,
Journal of Forensic Psychiatry & Psychology 23(5-6), 620-634.
Epstein, S., & Morling, B. (1995). Is the self motivated to do more than
enhance and/or verify itself? In M. H. Kernis (Ed.), Efficacy,
Agency, and Self-Esteem (pp. 9–30). New York: Plenum.
178
Falkenbach, D.M., Howe, J.R., & Falki, M. (2013). Using self-esteem to
disaggregate psychopathy, narcissism and aggression, Personality
& Individual Differences, 54, 815-820.
Farnham, S. D., Greenwald, A. G., & Banaji, M. R. (1999). Implicit self-
esteem. In D. Abrams,& M.A. Hogg (Eds.), Social Identity and
Social Cognition (pp. 230–248). London: Blackwell.
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A
flexible statistical power analysis program for the social,
behavioral, and biomedical sciences. Behavior Research Methods,
39, 175-191.
Fazel, S., & Danesh, J. (2002). Serious mental disorder in 23000
prisoners: a systematic review of 62 surveys, Lancet, 359(9306),
545-50.
Fiske, S. T., Gilbert, D. T., & Lindzey, G. (Eds.). (2010). The Handbook of
Social Psychology (Vol. 1, 5th ed.). Hoboken, NJ: Wiley.
Fonagy, P. (1999). Male perpetrators of violence against women: An
attachment theory perspective, Journal of Applied Psychoanalytic
Studies, 1, 7-27.
Fruehwald, S., et al. (1998). The relevance of self-concepts discriminating
in long-term incarcerated sex offenders, Journal of Behavior
Therapy and Experimental Psychiatry 29(4), 267-278.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L. (1998). Measuring
individual differences in implicit cognition: The Implicit Association
Test. Journal of Personality and Social Psychology, 74, 1464-
1480.
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and
Using the Implicit Association Test: I. An Improved Scoring
179
Algorithm. Journal of Personality and Social Psychology, 85, 197-
216.
Gregg, A. P., & Sedikides, C. (2010). Narcissistic fragility: Rethinking its
links to explicit and implicit self-esteem. Self and Identity, 9, 142-
161
Hare, R. D. (2003). Manual for the Revised Psychopathy Checklist (2nd
ed.). Toronto, ON, Canada: Multi-Health Systems.
Harpur, T.J., Hare, R.D., & Hakistan, A.R. (1989). Two factor
conceptualisation of psychopathy: construct validity and
assessment implications, Psychological Assessment: A Journal of
Consulting & Clinical Psychology, 1, 6-17.
Hetts, J., Sakuma, M., & Pelham, B. (1999). Two roads to positive regard:
Implicit and explicit self-evaluation and culture. Journal of
Experimental Social Psychology, 35, 512–559.
Hayes, A. F. (2012). PROCESS: A versatile computational tool for
observed variable mediation, moderation, and conditional process
modeling [White paper]. Retrieved from http://www.afhayes.com/
public/process2012.pdf
Kantor, G.K., & Jasinski, J.L.(1998). Dynamics and risk factors in partner
violence. In: Jasinski JL, Williams LM, (Eds). Partner Violence: A
Comprehensive Review of 20 Years of Research. Thousand Oaks,
CA, Sage.
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(9), 539-46.
Kernberg, O. F. (1975). Borderline Conditions and Pathological
Narcissism. Northvale, NJ: Jason Aronson, Inc.
180
Kernis, M. H. (2003). Toward a conceptualization of optimal self-esteem,
Psychological Inquiry, 14, 1-26.
Kernis, M. H., Abend, T., Goldman, B. M., Shrira, I., Paradise, A. W., &
Hampton, C. (2003). Discrepancies between explicit and implicit
self-esteem and self-serving responses, Self and Identity, 4, 311 –
330.
Kernis, M. H., Abend, T. A., Goldman, B. M., Shrira, I., Paradise, A. N., &
Hampton, C. (2005). Self-serving responses arising from
discrepancies between explicit and implicit self-esteem, Self and
Identity, 4(4), 311-330.
Kernis, M. H., Grannemann, B. D. & Barclay, L. C. (1989). Stability and
level of self-esteem as predictors of anger arousal and hostility.
Journal of Personaliry and Social Psychology, 56, 1013-1023.
Kesting, M.L., & Lincoln, T.M. (2013). The relevance of self-esteem and
self-schemas to persecutory delusions: a systematic review,
Comprehensive Psychiatry, 54 (7), 766-789.
Krause, S., Back, M. D., Egloff, B., & Schmukle, S. C. (2011). Reliability of
implicit self-esteem measures revisited. European Journal of
Personality, 25, 239-251.
Kroner, D. G.,&Weeks, J. R. (1996). Balanced inventory of desirable
responding: Factor structure, reliability, and validity with an
offender sample. Personality and Individual Differences, 21, 323-
333.
Lanyon, R.I., & Carle, A.C. (2007). Internal and external validity of scores
on the balanced inventory of desirable responding and the paulhus
deception scales, Educational and Psychological Measurement,
67(5), 1-19.
181
Laurell, J., Belfrage, H., & Hellstrom, A. (2010). Facets on the
psychopathy checklist screening version and instrumental violence
in forensic psychiatric patients, Criminal Behaviour and Mental
Health, 20(4), 285–294.
Lilienfeld, S.O. (1990). Development and Preliminary validation of a self-
report measure of psychopathic personality. Doctoral Dissertation,
University of Minnesota, Minneapolis.
Lilienfeld, S.O., & Andrews, B.P. (1996). Development and preliminary
validation of a self-report measure of measure of psychopathic
personality traits in noncriminal populations, Journal of Personality
Assessment, 66, 488-524.
Lilienfeld, S. O., & Hess, T. H. (2001). Psychopathic personality traits and
somatization: Sex differences and the mediating role of negative
emotionality, Journal of Psychopathology and Behavioral
Assessment, 23, 11–24.
Jordan, C.H., Spencer, S.J., & Zanna, M.P. (2005). Types of High Self-
Esteem and Prejudice: How Implicit Self-Esteem Relates to Ethnic
Discrimination Among High Explicit Self-Esteem Individuals,
Personality and Social Psychology Bulletin, 31(5), 693-702.
Jordan, C. H., Spencer, S. J., Zanna, M. P., Hoshino-Browne, E., & Correll,
J. (2003). Secure and defensive high self-esteem, Journal of
Personality and Social Psychology, 85(5), 969-978.
New, A.S. et al. (2004). Low prolactin response to fenfluramine in
impulsive aggression, Journal of Psychiatric Research, 38(3), 223-
230.
Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2007). The Implicit
Association Test at age 7: A methodological and conceptual
182
review (pp. 265–292). In J. A. Bargh (Ed.), Automatic Processes in
Social Thinking and Behavior. Psychology Press
Paulhus, D. L. (1991). Measurement and control of response bias. In J. P.
Robinson, P. R. Shaver & L. S. Wrightsman (Eds.), Measures of
Personality and Social Psychological Attitudes (pp. 17-59). San
Diego, CA: Academic Press.
Paulhus, D. L. (1994). Balanced Inventory of Desirable Responding:
Reference manual for BIDR Version 6. Unpublished manuscript, University
of British Columbia,
Vancouver, Canada.
Raskin, R., & Hall, C. S. (1979). A narcissistic personality inventory,
Psychological Reports, 45, 590.
Raskin, R. & Terry, H. (1988). A Principle-Components Analysis of the
Narcissistic Personality Inventory and Further Evidence of Its
Construct Validity, Journal of Personality and Social Psychology,
54, 890-902
Reidy, D.E., Foster, J.D., & Zeichner, A. (2010). Narcissism and
unprovoked aggression, Aggressive Behaviour, 36(6), 414-422.
Rosenberg, M. (1965). Society and the Adolescent Self-Image. Princeton,
NJ: Princeton University Press.
Sandoval, A.R., Hancock, D., Poythress, N., Edens, J.F., & Lilienfeld, S.
(2000). Construct validity of the psychopathic personality inventory
in a correctional sample, Journal of Personality Assessment, 74,
262-281.
Sarkar, J., & Duggan, C. (2010). Diagnosis and classification of personality
disorder: difficulties, their resolution and implications for practice,
Advances in Psychiatric Treatment, 16: 388-396.
183
Schneider, D. J., & Turkat, D. (1975). Self-presentation following success
or failure: Defensive self-esteem models. Journal of Personality,
43, 127–135.
Sellbom, M. (2011). Elaborating on the Construct Validity of the Levenson
Self-Report Psychopathy Scale in Incarcerated and Non-
Incarcerated Samples, Law & Human Behavior, 35, 440–451.
Smith, E. R., & DeCoster, J. (2001). Dual-process models in social and
cognitive psychology: Conceptual integration and links to
underlying memory systems, Personality and Social Psychological
Review, 4, 108–131.
Spalding, L. R., & Hardin, C. D. (1999). Unconscious unease and self-
handicapping: Behavioral consequences of individual differences
in implicit and explicit self-esteem. Psychological Science, 10,
535–539.
Stucke, T.S., & Sporer, S.L. (2002). When a grandiose self-image is
threatened: narcissism and self-concept clarity as predictors
of negative emotions and aggression following ego-threat,
Journal of Personality, 70(4), 509–532.
Whyte, S., Fox, S., & Coxell. A. (2006). Reporting of personality disorder
symptoms in a forensic inpatient sample: Effects of mode of
assessment and response style, The Journal of Forensic
Psychiatry & Psychology, 17(3), 431 – 441.
Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual
attitudes. Psychological Review, 107, 101–126.
Woodworth, M., & Porter, S. (2002). In cold blood: Characteristics of
criminal homicides as a function of psychopathy. Journal of
Abnormal Psychology, 111, 436-445.
184
Zeigler-Hill, V. (2006a). Discrepancies between implicit and explicit self-
esteem: Implications for narcissism and self-esteem instability.
Journal of Personality, 74, 119-143.
Zeigler-Hill, V., & Jordan, C.H. (2011). Behind the mask: Narcissism and
implicit self-esteem. In Campbell, W.K., & Miller, J.D. (Eds). The
Handbook of Narcissism and Narcissistic Personality Disorder:
Theoretical Approaches, Empirical Findings, and Treatments, UK:
John Wiley & Sons.
16.0 Appendices - measures
185
Appendix 1 – Rosenberg Self-Esteem Scale (RSES)Below is a list of statements dealing with your general feelings about yourself. If you strongly agree, circle SA. If you agree with the statement, circle A. If you disagree, circle D. If you strongly disagree, circle SD.
1. On the whole, I am satisfied with myself.
SA A D SD
2. At times, I think I am no good at all.
SA A D SD
3. I feel that I have a number of good qualities.
SA A D SD
4. I am able to do things as well as most other people.
SA A D SD
5. I feel I do not have much to be proud of.
SA A D SD
6. I certainly feel useless at times.
SA A D SD
7. I feel that I’m a person of worth, at least on an equal plane with others.
SA A D SD
8. I wish I could have more respect for myself.
SA A D SD
9. All in all, I am inclined to feel that I am a failure.
SA A D SD
10. I take a positive attitude toward myself.
SA A D SD
Appendix 2
Self Report Psychopathy Scale (SRPS)
186
Removed for copyright reasons
NPI
Removed for copyright reasons
Aggression: History
Number of previous physically aggressive convictions: ……………………………..
Counted by: ………………………………………………………………………………..
Source: …………………………………………………………………………………….
Index Offence Modified Overt Aggression Scale (MOAS)
Removed for copyright reasons
Institutional Aggression Modified Overt Aggression Scale (MOAS)
Removed for copyright reasons
CODING GUIDE FOR VIOLENT INCIDENTS:
INSTRUMENTAL VERSUS HOSTILE/REACTIVE AGGRESSION
187
Dewey G. Cornell
Curry School of Education
University of Virginia
This is the coding guide we used to code violent crimes as instrumental or hostile/reactive forms of aggression. The coding guide is being made available to researchers, but it is not an established clinical instrument and is intended only for
research purposes. For additional information, see the published study:
Cornell, D. G., Warren, J., Hawk, G., Stafford, E., Oram, G., & Pine, D. (1996). Psychopathy of instrumental and reactive violent offenders. Journal of Consulting
and Clinical Psychology, 64, 783-790.
October 4, 1996. These coding guidelines were developed for research purposes with grant support of the Harry Frank Guggenheim Foundation. Project researchers include Drs. Dewey Cornell, Gary Hawk, and Janet Warren. We thank Ed Stafford, Guy Oram, and Denise Pine for their contributions to this project. These guidelines
are subject to revision. Email [email protected].
188
CODING GUIDE FOR VIOLENT INCIDENTS
The primary distinction is between instrumental and reactive/hostile aggression. Originally we attempted to make this distinction through a global rating based on the rater's overall evaluation of the incident. However, some violent incidents had both instrumental and reactive/hostile qualities. For example, a person planned and carried out a robbery, but in the course of the robbery became angry when a storekeeper resisted him, and shot him in anger. Therefore, we decided to give priority to the presence of instrumental qualities, based on the theory that reactive hostility is the more common, pervasive form of aggression in criminal behavior and that instrumental aggression in criminal behavior represents a more pathological development and elaboration of the capacity for reactive aggression.
In addition to coding for the presence of instrumental and reactive aggression, the coders will make secondary ratings of these specific aspects of the aggressive act:
1) Planning - degree of premeditation or preparation for aggression
2) Goal-directedness - degree to which aggression is motivated by some external gain or incentive such as money
3) Provocation - degree of provocation, frustration or threat from victim
4) Arousal - degree of anger experienced by aggressor
5) Severity of violence - degree of injury to victim
6) Relationship to victim - closeness of relationship between victim and aggressor
7) Intoxication - intoxication on drugs or alcohol during incident
8) Psychosis - presence of psychotic symptoms during incident
These secondary ratings reflect aspects of the aggressive act which are not necessarily independent of one another. For example, planning and goal-directedness may be correlated. However, each of the components can be distinguished conceptually from the others and we are able to identify specific cases which support these distinctions.
In our discussion of various aggressive acts, the secondary ratings (especially the first four) seem to tap characteristics which contribute to the primary distinction between reactive and instrumental aggression, but these ratings are not equivalent to it. We used the secondary ratings to examine several questions:
1) Is there a stable combination or set of decision rules for the secondary ratings which is equivalent to the primary distinction?
2) Do the secondary ratings permit a sub-classification or refinement of the primary distinction which improves upon it?
189
Subjects may be dishonest, inaccurate, or incomplete in their account of the offense. Consider all available sources. Code what you believe to be true, what actually happened. If the subject claims self-defense, but all other available information indicates otherwise, and the subject is of doubtful credibility, code what you believe to be true.
Instrumental Aggression
The two cardinal characteristics of instrumental aggression are goal-directedness and planning. The instrumental aggressor acts to obtain a readily apparent goal such as power, money, sexual gratification, or some other objective beyond inflicting injury on the victim. Examples of instrumental aggression include shooting a police officer in the course of a bank robbery, stabbing a homeowner during a burglary, and strangling a rape victim. Rape is almost always instrumental. Sadistic aggression is a special form of instrumental aggression in which the objective is some form of pleasure (e.g., power or sexual gratification) that stems from the infliction of pain or attainment of dominance over the other person. Instrumental aggression is initiated as a means to an end rather than as an act of retaliation or self-defense.
Instrumental aggression often involves planning or preparation. However, in some cases instrumental aggression involves relatively little planning, such as in the case of a criminal who engages in an opportunistic offense (e.g., unexpected opportunity to rob someone that involves assaulting the victim). In some cases, a subject may plan a robbery or burglary, and when something goes wrong, engages in an act of aggression, such as shooting someone in order to get away. In these cases the coder should consider that the subject's plans included the possibility of violence, even if there was no specific plan to shoot someone.
Instrumental aggression usually involves little or no provocation by the victim. In some cases subjects may be "provoked" into violence in the course of another crime, e.g., a robbery victim who insults the subject or resists the robbery in some way. These acts are still considered instrumental acts of aggression.
Instrumental aggressors are motivated by goals, not emotions. It follows that their level of emotional arousal, especially anger, is relatively low or is secondary to the act. Some instrumental aggressors try to calm themselves prior to an offense through drug use or drinking. In extreme cases, instrumental aggressors are not angry toward their victims and may have a cold, "business-like" attitude about their behavior. Nevertheless, many less hardened instrumental aggressors are nervous and highly aroused while committing a crime, even though it is not their arousal which motivates their actions.
The term "instrumental" should not be defined so broadly that it encompasses all aggressive behavior simply because there is a definable goal or desired outcome to the aggression, such as warding off an attacker or taking revenge on someone. Aggressive behavior whose purpose is to
190
defend against a threat or in some way respond to provocation is defined as reactive/hostile aggression. If the subject is engaged in some form of criminal activity, such as a drug deal, associated violence is almost always instrumental.
Reactive/Hostile Aggression
The two cardinal characteristics of reactive/hostile aggression are reaction to provocation and arousal of hostility. Aggressive behavior represents reactive hostility to the extent that the aggressor reacts to perceived provocation or threat by the victim. The provocation may include insults, threats of aggression, or other acts that frustrate and anger the aggressor. The objective of the aggressive act is to harm or injure the victim, in response to feelings of hostility that may include a mixture of anger, resentment, fear, or other distress aroused by the victim's actions. Typically, there should be some form of interpersonal conflict (argument, dispute, prior aggression) between aggressor and victim. In many cases the aggressor and victim have a prior relationship as relatives or acquaintances, but in other cases there is no prior relationship and the parties are strangers to one another.
Bear in mind that reactive/hostile aggression can involve extended time-frames. For example, an abused family member may plan an ambush to rid the family of the abuser. The most recent episode of abuse could be long before the aggressive reaction. The critical issue is that the reactive/hostile subject is reacting to an interpersonal conflict that arouses hostility.
3 - Clearly instrumental aggression
2 - Both reactive and instrumental qualities are prominent
(subsequently these cases combined with instrumental group)
1 - Clearly reactive hostile aggression
Do not consider "displaced anger" or any form of displacement from one situation to the next. Many instrumental offenders may be angry at someone else, upset over a failed relationship, lost job, etc. This provides a context for understanding the person, but it should not enter into the determination that a person engaged in instrumental versus reactive/hostile violence. A person who sets out to rob a bank is committing an instrumental act, regardless of any prior life stress. A person who is embroiled in an intense interpersonal conflict with the victim will commit a reactive/hostile offense.
191
SECONDARY SCALES FOR VIOLENT INCIDENTS
Planning
How much did the subject plan or prepare for the aggressive action? Consider both the length of time involved in preparation and the amount of preparatory activity.
4 - extensive planning (detailed plan or preparation, rehearsal)
3 - moderate planning (contemplation of action for more than 24 hours)
2 - some planning (action within 24 hours, some plan or preparation)
1 - very little or no planning (acts during argument or fight, no preparation)
Assign a (1) to actions which are part of contiguous event, such as pausing during an argument to grab and load a gun. Assign a (2) if there is a break in the argument where the subject leaves the scene of an argument and returns with a gun later in the day.
Goal-Directedness
How much is the subject motivated by an external incentive, goal, or objective beyond just responding to provocation or threat? Readily apparent goals include money, power, sexual gratification, or some other external goal of benefit to the aggressor. Do not include such goals as self-defense, escaping harm, taking revenge for previous aggression, or acting out of frustration.
4 - Clear, unequivocal goal-directedness (include shooting during crimes)
3 - Primary goal-directedness, with presence of other motives
2 - Secondary goal-directedness, in presence of other primary motives
1 - No apparent goal-directedness (motive to injury victim, retaliate, defend)
Provocation
Did the victim's actions provoke the subject's aggression? Include provocation that occurred prior to the incident (e.g., prior abusive treatment).
6 - Exceptionally strong provocation (repeated assault, severe abuse)
5 - Very Strong provocation (assault)
4 - Strong (break-up of a romantic relationship, threat of major life change)
3 - Moderate provocation (serious argument or dispute, threat of assault)
2 - Mild provocation (insult, minor argument, confrontation with police)
1 - No apparent provocation
192
Consider the subject's personal point of view, even if the subject has a delusional perception of threat.
193
Arousal
How much emotional arousal, especially anger, did the subject experience at the time of the aggressive act? Just code the subject's mental state, not attitude toward the victim.
4 - Enraged, furious, described as "out of control" or "irrational"
3 - Angry, mad, extremely frightened (can be protracted state)
2 - Excited, very nervous, anxious
1 - Calm or tense at most
Arousal at the (4) level is extraordinary, and should be of short duration.
Severity of violence
7 - Extreme homicide (multiple killing, mutilation)
6 - Homicide
5 - Severe injury (lasting impairment or life-threatening injury, some rapes)
4 - Serious injury, requiring substantial hospital treatment (broken limb, rape,
gunshot)
3 - Minor injury (e.g., bruises, minor medical treatment, attempted rape)
2 - Assault without injury
1 - No assault (e.g., threatened with weapon)
Relationship with victim
Code the degree of contact or closeness between aggressor and victim. The scores listed here are typical scores. Some relationships may require higher or lower scores than indicated. Generally give maximum scores to immediate family members, unless there has been prolonged separation or lack of contact that substantially alters the relationship (e.g., father who never lived in the home, mother who turned over care of child to grandmother). A step-parent may receive the same score as a parent if there appears to have been similar bonding and contact since early childhood. Code based on duration and closeness of relationship.
5 - Very close relationship (immediate family member, romantic partner)
4 - Close relationship (friend, relative, dating partner, etc.)
3 - Specific relationship (teacher, babysitter, etc.)
2 - Acquaintance
1 - Stranger
Intoxication
194
Code whether the subject was intoxicated at the time of the aggressive incident. Consider alcohol and other drugs. Primary concern is degree to which the person is impaired or has clouded consciousness. Consider how much intoxication played a role in the subject's actions.
195
4 - Severe intoxication (large quantities of alcohol or drugs, very impaired)
3 - Intoxicated
2 - Mild intoxication (e.g., 1 or 2 drinks)
1 - Not intoxicated
Generally code (4) for subjects who are "falling down drunk" or extremely impaired by multiple substances, etc.
Psychosis (reality testing, not mood)
4 - Substantial psychotic symptoms (e.g., bizarre or pervasive delusions)
3 - Moderate psychotic symptoms (intermittent voices or delusions)
2 - Non-psychotic disturbance (e.g., depersonalized)
1 - Not psychotic
Generally code (4) for subjects who are very impaired by psychosis and have active symptoms. What you might call "falling down psychotic." Code (3) for individuals with mild, residual symptoms or more circumscribed symptoms that do not seriously impair everyday functioning. A man with a paranoid delusion about the victim who is nevertheless able to hold a job and function in many social situations is a (3). An actively psychotic man living on the street is probably a (4).
196
Subject Coder
VIOLENT INCIDENT CODING SHEET
Incident date:
Instrumental v Reactive/Hostile (code actual event, not just subject's claim)
4 - Clearly instrumental aggression (e.g., crime-related incident, drug deal)
3 - Primarily instrumental, some reactive qualities
2 - Primarily reactive hostile aggression, some instrumental qualities
1 - Clearly reactive hostile aggression (e.g., interpersonal conflict)
Planning (include plans for robbery, burglary, etc.)
4 - extensive planning (detailed plan or preparation, rehearsal)
3 - moderate planning (contemplation of action for more than 24 hours)
2 - some planning (action within 24 hours, some plan or preparation)
1 - very little or no planning (acts during argument or fight, no preparation)
Goal-Directedness (consider goals like financial gain, not just revenge)
4 - Clear, unequivocal goal-directedness (include shooting during crimes)
3 - Primary goal-directedness, with presence of other motives
2 - Secondary goal-directedness, in presence of other primary motives
1 - No apparent goal-directedness (motive to injure victim, retaliate, defend)
Provocation (includes provocation prior to incident, use subject's perception)
6 - Exceptionally strong provocation (repeated assault, severe abuse)
5 - Very Strong provocation (assault)
4 - Strong (break-up of a romantic relationship, threat of major life change)
3 - Moderate provocation (serious argument or dispute, threat of assault)
2 - Mild provocation (insult, minor argument, confrontation with police)
1 - No apparent provocation
Arousal (mental state, primarily code anger, but also consider other affects like fear)
197
4 - Enraged, furious, described as "out of control" or "irrational" or panicked (brief state)
3 - Angry, mad, extremely frightened (can be protracted state)
2 - Excited, very nervous, anxious, scared
1 - Calm or tense at most
Severity of violence (consider actual harm to victim, not subject's intention)
7 - Extreme homicide (multiple victims or multiple fatalities, mutilation)
6 - Homicide
5 - Severe injury (e.g., lasting impairment or life-threatening injury, some rapes)
4 - Serious injury, requiring substantial hospital treatment (e.g, broken limb, rape, gunshot)
3 - Minor injury (e.g., bruises, minor medical treatment, attempted rape)
2 - Assault without injury
1 - No assault (e.g., threatened with weapon)
Relationship with victim (if 2 or more victims, code highest)
5 - Very close relationship (immediate family member, romantic partner)
4 - Close relationship (friend, relative, dating partner, etc.)
3 - Specific relationship (teacher, babysitter, etc.) or Between friend and acquaintance
2 - Acquaintance
1 - Stranger Coding
198
Intoxication
4 - Severe intoxication (large quantities of alcohol or drugs, very impaired)
3 - Intoxicated
2 - Mild intoxication (e.g., 1 or 2 drinks)
1 - Not intoxicated
Psychosis (reality testing, not mood)
4 - Substantial psychotic symptoms (e.g., bizarre or pervasive delusions
3 - Moderate psychotic symptoms (intermittent voices or delusions)
2 - Non-psychotic disturbance (e.g., depersonalized)
1 - Not psychotic
199
CODER RELIABILITY STUDIES
We completed two reliability studies on the classification of instrumental and reactive offenders and the accompanying eight offense scales. In the first study, these scales were applied to a sample of 20 criminal defendants evaluated at the UVA Forensic Clinic. For five judges, the intraclass correlation coefficient for the scale distinguishing instrumental from reactive violence was .98. For 18 subjects all five judges agreed on violence type, and for the remaining two subjects four of five judges were in agreement. The other eight scales are listed below:
1) Planning (degree of planning and preparation prior to violence) .97;
2) Goal-directedness (presence of goals such as obtaining money) .94;
3) Provocation (subject's perception that victim provoked violence) .81;
4) Arousal (subject's degree of anger and excitement during violence) .83;
5) Severity of violence (degree of injury to victim) .97;
6) Relationship with victim (subject-victim relationship) .92;
7) Intoxication (alcohol or drug intoxication during violence) .96;
8) Psychosis (presence of psychotic symptoms during violence) .96.
In the second reliability study, we applied the slightly modified scales to records of 33 violent offenders incarcerated at the Staunton Correctional Center, a medium security state prison in Virginia. For 2 judges, the intraclass correlation for instrumental/reactive distinction was .93. The intraclass correlations for the other scales were all above .75, except for two scales: 1) the correlation for the provocation scale was .50, apparently because the records did not provide consistent information on the victim's behavior prior to the violent incident; and 2) the psychosis scale could not be used because none of the inmates were described as psychotic at the time of the offense.
Offense characteristics of instrumental and reactive violence. We examined the association between the instrumental/reactive classification and each of the eight offense variables for 50 Forensic Clinic defendants. These analyses were conducted for descriptive purposes to refine and clarify our conceptualization of instrumental and reactive aggression. Briefly, these analyses indicated that no single offense characteristic is synonymous with the instrumental/reactive distinction. The characteristics most strongly associated with instrumental violence are presence of a clearly definable goal, little or no provocation by the victim, and comparatively low levels of emotional arousal at the time of the offense. In contrast, reactive violence is associated most strongly with a lack of goal-directedness, little or no prior planning, provocation by the victim, and comparatively greater emotional arousal at the offense. Reactive violence more often involves family member victims, while instrumental violence is more often associated with acquaintances or strangers.
200
Categorical versus dimensional classification. We have found that specific violent incidents can be readily and reliably classified categorically as reactive or instrumental. Relatively few offenses pose classification difficulties, and in those cases we give greater weight to the presence of instrumental characteristics such as goal-directedness. However, the lifetime classification of violent offenders who have committed multiple offenses raises additional problems. Some offenders have extensive histories of reactive (or instrumental) violence while others have little or no history of violence prior to their recent offense. These cases suggest it may be viable to place subjects along a continuum for severity of reactive (or instrumental) violence.
Moreover, some offenders have a history of both reactive and instrumental violent offenses. (We have conducted detailed case studies of subjects with "mixed histories" of both instrumental and reactive violence, and it is clear that such individuals tend to be more similar to purely instrumental offenders, particularly in the presence of psychopathic characteristics.) A study by Vitiello, et al. (1990) found that violent juveniles fell into two groups, one an affective or reactive group, but the other a mixed group with both affective and predatory or instrumental aggression. Since persons can engage in both reactive and instrumental aggression, there is no reason why some offenders would not engage in both reactive and instrumental violence.
An alternative to the lifetime classification of offenders into instrumental, reactive, and mixed groups is to treat instrumental and reactive violence as separate dimensions. We are giving further consideration to this possibility. Our work linking instrumental violence to individual psychopathy is referenced below.
Cornell, D. G., Warren, J., Hawk, G., Stafford, E., Oram, G., & Pine, D. (1996). Psychopathy of instrumental and reactive violent offenders. Journal of Consulting and Clinical Psychology, 64, 783-790. Offense Characteristics of Reactive
and Instrumental Violent Offenders
Reactive
N=32
Instrumental
N=18
Planning
None
Some
Extensive
23
6
2
6
11
1
Goal Directedness
Clearly goal-directed
Mixed motives
No apparent goal-directedness
0
2
30
12
6
0
Provocation by Victim
Strong
Moderate
6
16
10
0
1
17
201
Weak or none
Arousal at Offense
Enraged
Angry
Excited/tense
Calm
5
21
3
1
0
1
9
4
Harm to Victim
Homicide
Serious injury
Not serious
24
6
2
8
4
6
Relationship to Victim
Very close (family)
Close (friend)
Acquaintance
Stranger
19
7
5
1
2
2
6
8
Intoxicated at Offense
Intoxicated
Some use
No use
12
2
18
9
0
9
Psychotic at Offense
Clearly psychotic
Some disturbance
Not psychotic
5
2
25
0
3
15
Note: Classification based on most recent offense.
Incomplete information on some variables.
202
Appendix 13
Participant Information Sheet
203
204
205
Appendix 14
Consent form
206
University of Surrey
PsychD Clinical Psychology
Major Research Project Proposal
URN: 6288605
The Relationship Between Self-Esteem Fragility and Aggression in a
Clinical Forensic Sample
Word count: 2962 (excluding title page, tables, references, and appendices)
207
1.0 Background and Theoretical Rationale
A recent literature review (see Appendix 1) concluded that disturbances in
three key aspects of self appear to be related to aggressive behaviour in
samples of people with personality disorder (PD) diagnoses and/or forensic
histories: self-esteem, self-concept and identity diffusion. The existing
evidence base appears to be compromised by poorly defined aspects of self,
making it difficult for one to identify exactly what is being measured, as
well as one-dimensional measures of rather complex, multi-dimensional
constructs. Therefore studies employing more rigorous, in-depth measures
of aspects of self are urgently required.
The role of the self in social and psychological functioning is well
documented (Campbell, Assanand & Di Paula, 2000). Self-disturbances are
prominent features of many mental health difficulties (Edwards et al., 2012;
Fonagy, 1999; Bateman & Fonagy, 2004), and have been highlighted as
important risk factors in violent and aggressive behaviour (Baumeister et al.,
1996; Donnellan et al., 2005; Kantor & Jasinski, 1998). In particular, self-
esteem is an important aspect of self that has been linked to aggression,
although whether it is high or low self-esteem is a contentious issue with
research finding that both low and high self-esteem are related to violence
(Baumeister, Smart & Boden, 1996; Baumeister, Bushman & Campbell,
2000; Boden, Fergusson, & Horwood, 2007). These conflicting findings
may be explained by the consideration of self-esteem as a unilateral
construct, using only explicit, self-report measures, which do not account
for the unconscious, implicit component of self-esteem. Dual processing
models demonstrate that information can be processed implicitly and
208
explicitly simultaneously, with the former being outside of conscious
awareness (Greenwald & Banaji, 1995). According to The Full Discrepancy
Model (Figure 1), dissonance between implicit and explicit self-esteem
generates a tenuous evaluation of self-worth, and it is this discrepancy that
is an important mediator of aggressive behaviour (Kernis, 2003; Kernis,
2005; Edwards and Bond, 2012).
The Partial-Discrepancy model emphasises the
low implicit/high explicit self-esteem cell (top
right of Figure 1) in people with high levels of
narcissistic traits and suggests that this
particular discrepancy is linked to higher levels
of aggressive behaviour. An explanation for
this relationship is that people with discrepant self-esteem characterised by
high explicit and low implicit attitudes (fragile self-esteem) towards the self
lack a secure, intrinsic base for their positive self-evaluations. Therefore
they react strongly to negative feedback and use maladaptive defences such
as aggression to attack the perceived source of the threat in order to defend
their positive, but vulnerable, feelings of self-worth (Kernis, Granneman et
al., 1989). This pattern of explicit, grandiose evaluation of the self
concealing unconscious negative feelings of self-worth are consistent with
current models of narcissism (Campbell & Foster, 2007; Zeigler-Hill,
2006a) and psychopathy (Cleckley, 1941). Little is known about the impact
of high implicit, low explicit (damaged) self-esteem of aggressive
behaviour, and research is needed to explore the direction of the discrepancy
between implicit and explicit self-esteem.
209
Most of the research into the relationship between fragile self-esteem and
aggression come from studies of non-clinical and non-forensic populations.
Despite the developments in our understanding of the importance of
implicit, as well as explicit processes involved in self-esteem, research in
clinical and forensic psychiatric settings are still using uni-dimensional,
self-report measures of self-esteem. This is surprising as self-esteem
fragility is a central component of narcissism and psychopathy (APA, 2013;
Cleckley, 1941; Hare, 2003), both of which are consistently associated with
aggression in research (for reviews see Reidy, Shelley-Tremblay &
Lilienfeld, 2011; Roberts & Coid, 2007 & Baumeister et al, 2000).
Therefore research is needed to explore whether self-esteem fragility
mediates the relationship between narcissism and psychopathy and
aggression in forensic psychiatric samples.
There is a high prevalence of people with diagnoses that are characterised
by an enduring, fragile sense of self-worth (such as Narcissistic PD and
Psychopathy) in forensic populations (Fazel & Danesh, 2002; New et al.,
2004), so it is possible that clinical levels of self-disturbance are related to
an increased risk of violence behaviour. Psychopathy and narcissism share a
number of traits including a lack of empathy, grandiosity, and exploitation
of others, with some researchers describing psychopathy as an extreme form
of pathological narcissism, and a type of narcissism particularly related to
violence (Kernberg, 1975; Cale & Lilienfeld, 2006). It is therefore possible
that self-esteem fragility mediates the relationship between narcissism,
psychopathy and aggression. There are longstanding debates regarding the
validity and reliability of categorical PD diagnoses in forensic populations
210
(Sarkar & Duggan, 2010), therefore a more suitable approach may be to
examine underlying traits of personality factors linked to aggression.
Findings could be used to predict recidivism and future risk of violence
therefore research is urgently needed in this area. These findings also have
important implications for the treatment of offenders, suggesting that, rather
than delivering treatment programs that target the level of offenders’ self-
esteem, treatment plans should be developed aiming to enhance congruence
between conscious and unconscious feelings of self-worth.
1.2 Main Hypotheses
5. Aggression (higher severity of index offence violence, greater of
violent crimes committed, higher number and greater severity of
aggressive incidents on the ward in the past month) will increase as
implicit self-esteem decreases, but only when explicit self-esteem is
high (this will represent self-esteem fragility)
6. Self-esteem fragility will predict higher levels of narcissism and
psychopathy
7. Self-esteem fragility will mediate the effect of narcissism on
aggression
8. Self-esteem fragility will mediate the effect of psychopathy on
aggression
2.0 Method
2.1 Design
211
A correlational design will be used. The relationships between personality
constructs (narcissism, psychopathy, implicit and explicit self-esteem) and
aggression will be explored using correlational and regression analyses.
2.2 Participants
A purposive sample of forensic psychiatric participants will be recruited
from <name of hospital removed> using the criteria outlined in Table 1.
This will be assessed by discussing with clients’ Responsible Clinician
(Consultant Psychiatrist), and confirmed by file review. Active grandiose
delusions will be an exclusion criteria because they are thought to be a
defence serving the function of maintaining self-esteem (Bentall, 1994).
Therefore delusions may reflect a temporary change in self-esteem rather
than the nature of evaluations of self-worth one would expect to see in
people with high levels of psychopathic and narcissistic traits.
Table 1. Inclusion and Exclusion Criteria
Inclusion criteria Exclusion criteria Fluent in English Sufficiently mentally stable
to participate and able to give informed consent (this will be checked with the nurse in charge on the day)
Violent forensic history
Developmental disorder or traumatic brain injury
Active grandiose delusions (i.e. fixed beliefs)
Previous studies looking at the effect of self-esteem on aggressive behaviour
have presented effect sizes of r=-0.31 in a forensic sample (Fruehwald et al.
1996) and r=-0.36 in an undergraduate sample (Falkenbach et al, 2013).
Assuming a power of 0.8 to detect an effect size of 0.335 (average of the
two effect sizes), one-tailed with alpha=0.05 using a correlational test, an a
212
priori calculation using G*Power 3.1.7 (Faul, Erdfelder, Lang & Buchner,
2007) suggested a sample size of 51 is needed.
Data from <name of hospital removed> indicates that there are 150 people
who may meet the inclusion criteria for this study. A recent study conducted
with a similar sample investigating a similar topic (Edwards & Bond, 2012)
reported that 55% of eligible participants agreed to participate. So if 55% of
the available population decided to take part that would yield a sample of 83
participants. Therefore, a sufficient number of participants are likely to take
part to meet minimum sample size recommended by the power calculation. I
will aim to recruit 60 participants.
2.3 Measures
Explicit Self-Esteem
The Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) consists of 10
self-report questions measuring global self-esteem with a maximum score of
30. A higher score indicates higher self-esteem. The scale has been
validated for use with a variety of clinical groups, with satisfactory
reliability and validity reported in a prison sample (Cale & Lilienfeld,
2006). I have chosen to use this measure as it is the most widely used
measure of explicit self-esteem and would allow me to compare my findings
with those from other studies in this area. A copy of the measure can be
found in Appendix 2.
Implicit Self-Esteem
The Implicit Association Test (IAT, Greenwald, McGhee & Schwartz,
1998) will be used to measure implicit self-esteem. It is a computerised task
213
that measures unconscious associations between self-relevant and non-self-
relevant words with pleasant and unpleasant words. The task will be run on
an NHS laptop meeting the necessary security requirements of <name of
hospital removed>. For a full description and instructions see Appendix 3)
The IAT has been chosen as the measure of implicit self-esteem because it
is the most widely used measure and it has shown the highest reliability of
all the available measures of implicit self-esteem (Bosson et al, 2000;
Krause et al. 2011). Scoring will be based on the IAT d calculation
described in Greenwald, Nosek and Banaji (2003). Negative values of d
signify low implicit self-esteem whereas high values indicate high implicit
self-esteem.
The words used by Gregg & Sedikides (2010, Table 2 in Appendix 3) may
not be appropriate for participants at <name of hospital removed> because
they are likely to strongly associate “self” words to words such as “murder”
due to their forensic history rather than because of low-implicit self-esteem.
Therefore the pleasant and unpleasant words will be updated using words
from The Affective Norms for English Words (ANEW; Bradley & Lang,
2010), ensuring that word frequency in everyday language, valence, arousal
and length of words are balanced between conditions.
Fragile Self-Esteem
The method for calculating the discrepancy between implicit and explicit
self-esteem, as well as the direction of the discrepancy is described in the
Data Analysis section.
Psychopathy
214
I have chosen to examine underlying traits of psychopathy and narcissism
rather than select a sample of people with related diagnoses because of the
unreliability and difficulty with diagnosing these disorders in forensic
populations (Sarkar & Duggan, 2010). The Levenson Self-Report
Psychopathy Scale (SRPS; Levenson et al, 1995) is a 26 item self-report
measure which is designed to assess personality and behavioural traits
associated with psychopathy. Participants are required to rate the extent to
which they agree or disagree with statements using a four point scale (see
Appendix 4). Higher scores are associated with higher levels of
psychopathy. This measure constitutes two factors; Primary Psychopathy
which measures a manipulative, callous interpersonal style and Secondary
Psychopathy which assesses behavioural aspects of psychopathy such as
inability to learn from mistakes and impulsivity. These two factors
correspond approximately to the two factors of the Hare Psychopathy
Checklist (PCL-R; Harpur, Hare & Hakistan, 1989). The SRPS was chosen
to measure psychopathy because it is the least time consuming for
participants compared to other lengthy scales, yet retains good convergent
and discriminative validity (Sellbom, 2011). Levenson et al. (1995) found
good internal consistency for the scale.
Narcissism
The Narcissistic Personality Inventory (NPI) is a 40 item self-report scale
designed to assess narcissistic traits (Raskin & Hall, 1979). This scale is
widely used in social science and personality research and has good internal
consistency (Raskin & Terry, 1988). NPI scores correlate significantly with
215
observer and self-report measures of narcissism (Raskin & Terry, 1988),
indicating satisfactory validity. For a copy of the measure see Appendix 5.
Outcome Measures: Aggression
All measures will relate to direct, physical aggression only. The number of
previous physically aggressive convictions, and the number of physically
violent incidents on the ward over the past 12 months will be counted. This
information will be obtained from clients’ medical files.
The Modified Overt Aggression Scale (MOAS; Kay, Wolkenfeld & Murrill,
1988) assesses four types of aggressive behaviour; verbal aggression,
aggression against property, auto-aggression and physical aggression over
the past week. Items are scored on a five point scale with high scores
indicating more aggression (see Appendix 6). This scale will be used to rate
the severity of violence of the index offence as well as the degree of
violence of incidents on the ward. Kay et al. (1988) reported good inter-rater
reliability (r=0.94), and good discriminative validity for distinguishing high,
intermediate and low violent groups. Twenty percent each of scored MOAS
forms for index offence and incidents on the ward will be randomly selected
and passed to Simon Draycott to rate to establish the degree of inter-rater
reliability.
2.4 Procedure
216
/
2.5 Ethical Considerations
All data will be stored securely on NHS computers, encrypted laptops or
encrypted USB sticks. All data entry and analysis will be performed by
myself. My supervisors Erica Hepper and Simon Draycott will have access
to anonymised data.
This study will involve some minor deception regarding the function of the
IAT. It is important that participants are not aware that it is a measure of
implicit self-esteem as this knowledge is likely to affect the way they
217
respond during the task, and their responses may not reflect their implicit
self-esteem. The task will therefore be called a measure of perception.
Participants will be debriefed after all data has been collected because if
they are told the function of the IAT straight after participating, they may
tell other participants which might affect their responses.
Some of the unpleasant, highly arousing words on IAT may distress
participants. This will be discussed with the service user forum at <name of
hospital removed>, and their opinion will be sought regarding whether
participants may find them too distressing, and whether any adaptations
need to be made.
This research proposal will be submitted to the local NHS ethics committee
(which will be allocated by a centralised process when the Research Ethics
Committee form is submitted) for a full review.
2.6 R & D Considerations
A research proposal was sent to the hospital clinical research panel on 18th
August as per the <name of hospital removed> research procedure, where is
it awaiting consideration by the panel. If it is approved, I will start preparing
my ethical approval application.
3.0 Project Costing
Each participant will be paid £3. 60*3 = £180
The money will be paid electronically into participants’ hospital account, as
patients are not allowed to have cash at <name of hospital removed>.
£20 to print information sheets and packs of questionnaires.
218
Total = £200
4.0 Proposed Data Analysis
In the first instance, correlational analyses will be performed to explore
relationships between all variables. Moderation and mediation analyses will
then be performed to test the hypotheses as outlined below:
Moderation Analyses
4. The main effects for implicit (IAT scores) and explicit self-esteem
(RSES scores) on violent history ratings (i.e. severity of index
offence and frequency and severity of previous violent convictions
scored using the MOAS) and institutional violence ratings (number
of violent incidents on the ward in the past month and severity
scored using the MOAS) will be investigated using regression
analyses.
5. Then the interaction of implicit and explicit self-esteem will be
entered into the regression model.
6. Follow up simple slopes tests (Aiken & West, 1991) will be
performed to explore the direction of the explicit and implicit self-
esteem interaction.
This will allow me to investigate Hypothesis One; that increased aggression
will increase as self-esteem decreases, but only when explicit self-esteem is
high (representing self-esteem fragility). This is the most common method
used in studies exploring the fragility of self-esteem (Jordan, Spencer, &
Zanna, 2005; Jordan, Spencer, Zanna, Hoshino-Browne, & Correll, 2003;
Kernis et al., 2005).
219
The analyses above will be repeated to investigate the extent to which the
interaction between explicit and implicit self-esteem predict scores on the
PCL-R and NPI. This will allow me to test Hypothesis Two; that self-esteem
fragility will predict higher levels of narcissism and psychopathy.
Mediation Analyses
A freely available computational tool for SPSS; The Hayes PROCESS
method (Hayes, 2012), will be used to explore the mediating effect of self-
esteem fragility on narcissism to predict aggression. This would allow me to
test Hypothesis Three; that esteem fragility will mediate the effect of
narcissism on aggression.
The same method will be used to explore the mediating effect of self-esteem
fragility on psychopathy to predict aggression, allowing me to test
Hypothesis Four, that self-esteem fragility will mediate the effect of
psychopathy on aggression.
5.0 Involving/Consulting Interested Parties
I will meet with representatives from the service user forum at
<name of hospital removed> for feedback regarding the words used
in the IAT.
A research proposal was submitted to <name of trust removed>
clinical research panel on 18/08/14 for review where it will be
decided whether approval will be given for me to conduct this study
at <name of hospital removed>. It is expected that the project will be
discussed at the panel on 06/10/14, with a decision being made
shortly after.
220
6.0 Contingency Plan
1. In the event that insufficient numbers are recruited from <name of
hospital removed>, I will source more participants from other medium-
secure forensic services within <name of areas removed>, within which my
supervisor (Simon Draycott) has contacts (for a list, see Appendix 7). This
would require a major amendment on the NHS ethical approval form.
2. If I am unable to conduct this study in <name of hospital removed> and I
am unable to access participants in the other MSUs detailed in Appendix 7,
a further contingency plan would be to use a non-clinical sample of students
and the general population, advertising the study and asking them to
complete the measures online, allowing collection of large amounts of data
in a relatively short period of time. The aggression measures will be
replaced with The Aggression Questionnaire (AQ; Buss & Perry, 1992), a
self-report measure of various types of aggressive behaviour. Ethical
approval would be required from the University of Surrey FAHS
department.
7.0 Dissemination Strategy
A summary sheet of the findings and implications from this study will be
sent to all participants. Results from this study will be disseminated locally
through the intranet site. The findings will be presented at local meetings at
all sites (<name of trust removed>, University of Surrey) and at the <name
of trust removed> local service user and carer forums. Findings will be
presented in oral and poster presentations at relevant conferences. They will
also be submitted for publication in peer-reviewed journals. The findings of
221
this study may serve as pilot data for further large-scale studies to be
completed (i.e. HTA funded), and subsequently may inform treatment and
risk management strategies for people with mental health difficulties and
forensic histories.
References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Newbury Park, CA: Sage Publications.
American Psychiatric Association, (2013). Diagnostic and Statistical
Manual of Mental Disorders (5th ed). Arlington, VA: American
Psychiatric Publishing.
Bateman, A.W., & Fonagy, P. (2004). Mentalization-based treatment of
BPD, Journal of Personality Disorders, 18, 36-51
Baumeister, R. F., Bushman, B. J., & Campbell, W. (2000). Self-esteem,
narcissism, and aggression: Does violence result from low self-
esteem or from threatened egotism? Current Directions in
Psychological Science 9(1), 26-29.
Baumeister, R. F., Smart, L. & Boden, J.M. (1996). Relation of threatened
egotism to violence and aggression: The dark side of high self-
esteem, Psychological Review 103(1), 5-33.
Baumeister, R.F., Smart, L., & Boden, J.M. (1996). Relation of threatened
egotism to violence and aggression: The dark side of high self-
esteem, Psychological Review, 103(1), 5-33.
222
Bentall RP, Kinderman P, Kaney S (1994) The self, attributional processes
and abnormal beliefs: Towards a model of persecutory delusions,
Behaviour, Research & Therapy, 32, 331–341.
Blackburn, R & Glasgow, D. (2006). Manual for the Chart of Interpersonal
Reactions in Closed Living Environments (CIRCLE). University of
Liverpool.
Boden, J. M., Fergusson, D.M., & Horwood, L. (2007). Self-esteem and
violence: Testing links between adolescent self-esteem and later
hostility and violent behaviour, Social Psychiatry and Psychiatric
Epidemiology 42(11), 881-891.
Bradley, M.M. & Lang, P.J. (2010). Affective Norms for English Words
(ANEW): Instruction manual and affective ratings. Technical Report
C-2. University of Florida, Gainesville, FL.
Buss, A.H., & Perry, M. (1992). The aggression questionnaire, Journal of
Personality & Social Psychology, 63(3), 452-459.
Cale, E. M. & Lilienfeld, S.O. (2006). Psychopathy factors and risk for
aggressive behavior: A test of the "threatened egotism" hypothesis,
Law and Human Behavior 30(1), 51-74.
Campbell, J.D., Assand, S., & Di Paula, A. (2000). Structural features of the
self-concept and adjustment, In A. Tesser, R.B. Felson, & J.M. Suls,
(Eds). Psychological Perspectives on Self and Identity (pp 115-138).
Washington, DC, US: American Psychological Association.
Campbell, K.W., & Foster, J.D. (2007). The narcissistic self: Background,
an extended agency model, and ongoing controversies. In C.
223
Sedikides & S.J. Spencer (Eds) Frontiers in Social Psychology: The
Self (pp 115-138). UK: Psychology Press.
Cleckley, H. (1941) The So-called Psychopathic Personality, With Special
Emphasis on his Status in the Selective Service, Journal of the
Medical Association of Georgia, 30, 466–472.
Cornell, D. G., Warren, J., Hawk, G., Stafford, E., Oram, G., & Pine, D.
(1996). Psychopathy of instrumental and reactive violent offenders.
Journal of Consulting and Clinical Psychology, 64, 783-790.
Dolan, M., & Blackburn, R. (2006). Interpersonal factors as predictors of
disciplinary infractions in incarcerated personality disordered
offenders, Personality and Individual Differences, 40, 897–907
Donnellan, M.B., Trzesniewski, K.H., Robins, R.W., Moffitt, T.E., & Caspi,
A. (2005). Low Self-Esteem Is Related to Aggression, Antisocial
Behavior, and Delinquency, Psychological Science, 16(4) ,328-335.
Edwards, R. & A. J. Bond (2012). Narcissism, self-concept clarity and
aggressive cognitive bias amongst mentally disordered offenders,
Journal of Forensic Psychiatry & Psychology 23(5-6), 620-634.
Epstein, S., & Morling, B. (1995). Is the self motivated to do more than
enhance and/or verify itself? In M. H. Kernis (Ed.), Efficacy,
Agency, and Self-Esteem (pp. 9–30). New York: Plenum.
Falkenbach, D.M., Howe, J.R., & Falki, M. (2013). Using self-esteem to
disaggregate psychopathy, narcissism and aggression, Personality &
Individual Differences, 54, 815-820.
224
Farnham, S. D., Greenwald, A. G., & Banaji, M. R. (1999). Implicit self-
esteem. In D. Abrams,& M.A. Hogg (Eds.), Social Identity and
Social Cognition (pp. 230–248). London: Blackwell.
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A
flexible statistical power analysis program for the social, behavioral,
and biomedical sciences. Behavior Research Methods, 39, 175-191.
Fazel, S., & Danesh, J. (2002). Serious mental disorder in 23000 prisoners:
a systematic review of 62 surveys, Lancet, 359(9306), 545-50.
Fiske, S. T., Gilbert, D. T., & Lindzey, G. (Eds.). (2010). The Handbook of
Social Psychology (Vol. 1, 5th ed.). Hoboken, NJ: Wiley.
Fonagy, P. (1999). Male perpetrators of violence against women: An
attachment theory perspective, Journal of Applied Psychoanalytic
Studies, 1, 7-27.
Fruehwald, S., et al. (1998). The relevance of self-concepts discriminating
in long-term incarcerated sex offenders, Journal of Behavior
Therapy and Experimental Psychiatry 29(4), 267-278.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L. (1998). Measuring
individual differences in implicit cognition: The Implicit Association
Test. Journal of Personality and Social Psychology, 74, 1464-1480.
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding
and Using the Implicit Association Test: I. An Improved Scoring
Algorithm. Journal of Personality and Social Psychology, 85, 197-
216.
225
Gregg, A. P., & Sedikides, C. (2010). Narcissistic fragility: Rethinking its
links to explicit and implicit self-esteem, Self and Identity, 9, 142-
161
Hare, R. D. (2003). Manual for the Revised Psychopathy Checklist (2nd
ed.). Toronto, ON, Canada: Multi-Health Systems.
Harpur, T.J., Hare, R.D., & Hakistan, A.R. (1989). Two factor
conceptualisation of psychopathy: construct validity and assessment
implications, Psychological Assessment: A Journal of Consulting &
Clinical Psychology, 1, 6-17.
Hetts, J., Sakuma, M., & Pelham, B. (1999). Two roads to positive regard:
Implicit and explicit self-evaluation and culture. Journal of
Experimental Social Psychology, 35, 512–559.
Hayes, A. F. (2012). PROCESS: A versatile computational tool for
observed variable mediation, moderation, and conditional process
modeling [White paper]. Retrieved from http://www.afhayes.com/
public/process2012.pdf
Kantor, G.K., & Jasinski, J.L.(1998). Dynamics and risk factors in partner
violence. In: Jasinski JL, Williams LM, (Eds). Partner Violence: A
Comprehensive Review of 20 Years of Research. Thousand Oaks,
CA, Sage.
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(9), 539-46.
Kernberg, O. F. (1975). Borderline Conditions and Pathological
Narcissism. Northvale, NJ: Jason Aronson, Inc.
226
Kernis, M. H. (2003). Toward a conceptualization of optimal self-esteem,
Psychological Inquiry, 14, 1-26.
Kernis, M. H., Abend, T., Goldman, B. M., Shrira, I., Paradise, A. W., &
Hampton, C. (2003). Discrepancies between explicit and implicit
self-esteem and self-serving responses, Self and Identity, 4, 311 –
330.
Kernis, M. H., Abend, T. A., Goldman, B. M., Shrira, I., Paradise, A. N., &
Hampton, C. (2005). Self-serving responses arising from
discrepancies between explicit and implicit self-esteem, Self and
Identity, 4(4), 311-330.
Kernis, M. H., Grannemann, B. D. & Barclay, L. C. (1989). Stability and
level of self-esteem as predictors of anger arousal and hostility.
Journal of Personaliry and Social Psychology, 56, 1013-1023.
Krause, S., Back, M. D., Egloff, B., & Schmukle, S. C. (2011). Reliability
of implicit self-esteem measures revisited. European Journal of
Personality, 25, 239-251.
Kroner, D. G.,&Weeks, J. R. (1996). Balanced inventory of desirable
responding: Factor structure, reliability, and validity with an
offender sample. Personality and Individual Differences, 21, 323-
333.
Lanyon, R.I., & Carle, A.C. (2007). Internal and external validity of scores
on the balanced inventory of desirable responding and the paulhus
deception scales, Educational and Psychological Measurement,
67(5), 1-19.
227
Laurell, J., Belfrage, H., & Hellstrom, A. (2010). Facets on the psychopathy
checklist screening version and instrumental violence in forensic
psychiatric patients, Criminal Behaviour and Mental Health, 20(4),
285–294.
Lilienfeld, S.O. (1990). Development and Preliminary validation of a self-
report measure of psychopathic personality. Doctoral Dissertation,
University of Minnesota, Minneapolis.
Lilienfeld, S.O., & Andrews, B.P. (1996). Development and preliminary
validation of a self-report measure of measure of psychopathic
personality traits in noncriminal populations, Journal of Personality
Assessment, 66, 488-524.
Lilienfeld, S. O., & Hess, T. H. (2001). Psychopathic personality traits and
somatization: Sex differences and the mediating role of negative
emotionality, Journal of Psychopathology and Behavioral
Assessment, 23, 11–24.
Jordan, C.H., Spencer, S.J., & Zanna, M.P. (2005). Types of High Self-
Esteem and Prejudice: How Implicit Self-Esteem Relates to Ethnic
Discrimination Among High Explicit Self-Esteem Individuals,
Personality and Social Psychology Bulletin, 31(5), 693-702.
Jordan, C. H., Spencer, S. J., Zanna, M. P., Hoshino-Browne, E., & Correll,
J. (2003). Secure and defensive high self-esteem, Journal of
Personality and Social Psychology, 85(5), 969-978.
New, A.S. et al. (2004). Low prolactin response to fenfluramine in
impulsive aggression, Journal of Psychiatric Research, 38(3), 223-
230.
228
Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2007). The Implicit
Association Test at age 7: A methodological and conceptual review
(pp. 265–292). In J. A. Bargh (Ed.), Automatic Processes in Social
Thinking and Behavior. Psychology Press
Paulhus, D. L. (1991). Measurement and control of response bias. In J. P.
Robinson, P. R. Shaver & L. S. Wrightsman (Eds.), Measures of
Personality and Social Psychological Attitudes (pp. 17-59). San
Diego, CA: Academic Press.
Paulhus, D. L. (1994). Balanced Inventory of Desirable Responding:
Reference manual for BIDR Version 6. Unpublished manuscript,
University of British Columbia, Vancouver, Canada.
Raskin, R., & Hall, C. S. (1979). A narcissistic personality inventory,
Psychological Reports, 45, 590.
Raskin, R. & Terry, H. (1988). A Principle-Components Analysis of the
Narcissistic Personality Inventory and Further Evidence of Its
Construct Validity, Journal of Personality and Social Psychology,
54, 890-902
Reidy, D.E., Foster, J.D., & Zeichner, A. (2010). Narcissism and
unprovoked aggression, Aggressive Behaviour, 36(6), 414-422.
Rosenberg, M. (1965). Society and the Adolescent Self-Image. Princeton,
NJ: Princeton University Press.
Sandoval, A.R., Hancock, D., Poythress, N., Edens, J.F., & Lilienfeld, S.
(2000). Construct validity of the psychopathic personality inventory
in a correctional sample, Journal of Personality Assessment, 74,
262-281.
229
Sarkar, J., & Duggan, C. (2010). Diagnosis and classification of personality
disorder: difficulties, their resolution and implications for practice,
Advances in Psychiatric Treatment, 16: 388-396.
Schneider, D. J., & Turkat, D. (1975). Self-presentation following success
or failure: Defensive self-esteem models. Journal of Personality, 43,
127–135.
Sellbom, M. (2011). Elaborating on the Construct Validity of the Levenson
Self-Report Psychopathy Scale in Incarcerated and Non-Incarcerated
Samples, Law & Human Behavior, 35, 440–451.
Smith, E. R., & DeCoster, J. (2001). Dual-process models in social and
cognitive psychology: Conceptual integration and links to
underlying memory systems, Personality and Social Psychological
Review, 4, 108–131.
Spalding, L. R., & Hardin, C. D. (1999). Unconscious unease and self-
handicapping: Behavioral consequences of individual differences in
implicit and explicit self-esteem. Psychological Science, 10, 535–
539.
Stucke, T.S., & Sporer, S.L. (2002). When a grandiose self-image is
threatened: narcissism and self-concept clarity as predictors of
negative emotions and aggression following ego-threat, Journal of
Personality, 70(4), 509–532.
Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual
attitudes. Psychological Review, 107, 101–126.
230
Woodworth, M., & Porter, S. (2002). In cold blood:
Characteristics of criminal homicides as a function of
psychopathy. Journal of Abnormal Psychology, 111, 436-445.
Zeigler-Hill, V. (2006a). Discrepancies between implicit and
explicit self-esteem: Implications for narcissism and self-
esteem instability. Journal of Personality, 74, 119-143.
Zeigler-Hill, V., & Jordan, C.H. (2011). Behind the mask: Narcissism and
implicit self-esteem. In Campbell, W.K., & Miller, J.D. (Eds). The
Handbook of Narcissism and Narcissistic Personality Disorder:
Theoretical Approaches, Empirical Findings, and Treatments, UK:
John Wiley & Sons.
231
Gantt Chart of Study Timeline
Sept 14
Oct Nov Dec Jan 15 Feb Mar April May June July Aug Sept Oct Nov Dec Jan 16
MRP Proposal submitted
Proposal passed (allowing for resubmission if required)IRAS and ethical approval preparation
Ethics submitted
Ethical approval received (allowing for amendments)R & D approval received
Induction (for honorary contract)
Data collection
Data Analysis
First draft submitted to supervisor
Complete draft submitted
University of Surrey
PsychD Clinical Psychology
Major Research Project Literature Review
URN: 6288605
Aggression and Aspects of the Self in Personality Disorders and Forensic
Populations: A Systematic Review of the Literature
Word count: 7877 (excluding title page, tables and references)
Abstract
Aim: The purpose of this study is to systematically review empirical
research investigating relationships between aspects of the self and direct
aggression in samples of people with personality disorders (PDs) and/or a
violent forensic background. Background: Further research into factors
associated with violence is urgently required due to the huge economic and
personal cost of this behaviour. Method: A literature search was undertaken
in Ovid and Web of Knowledge, then articles were screened for eligibility
using the inclusion criteria. Following deeper scrutiny, 19 articles were
selected for inclusion in the review, and were quality assessed. Results:
Results identified four key aspects of self that are associated with violence.
In PD populations, the small amount of research conducted indicated that a
range of impairments in aspects of self were associated with aggression, and
narcissism was found to be associated with the relationship between violent
behaviour and self-esteem. In forensic populations, self-concept and self-
esteem were found to be related to aggression, but the direction of the
relationship varied across studies. Conclusion: The findings suggest that
impaired, unstable aspects of self are related to aggression, and that they
may play a role in how offenders and people with PD diagnoses perceive
their behaviour. It appears that there are other important variables that
moderate this relationship. Limitations of the current evidence base and of
the present review are considered and possible explanations for the
variability of the findings are discussed. Clinical and theoretical
implications of the findings are highlighted, as well as possible directions
for future research.
233
Statement of Journal Choice
I have chosen to submit this systematic review to “Criminal Behaviour and
Mental Health”. I believe this review fits the scope of the journal, which
publishes papers that focus on the relationship between mental mechanisms
associated with criminal behaviour. This review focuses on aspects of the
self which play an important role in motivation, cognition and affect, and
the relationship they have to violence and aggression, which has vital
implications for criminal behaviour, and this journal represents a suitable
platform from which I can reach practitioners working in both fields. As this
paper views literature from forensic, as well as clinical populations, I have
chosen a journal that has no specific forensic or mental health focus. The
journal accepts systematic reviews. The review would be of interest and
relevance to the audience consists of clinical and non-clinical practitioners
working with offenders with and without mental health problems.
234
Introduction
Violent and aggressive behaviour refers to acts performed with the intention
of harming a victim. This behaviour creates significant economic and social
burdens on society and causes emotional and physical damage (Short et al.,
2012). For example violence has been reported as costing the British
economy £124 billion per year, which is equal to 7.7% of the UK’s Gross
Domestic Product (The Institute for Economics & Peace, 2013). Further, 3.7
million offences were recorded in the year ending June 2013 in England and
Wales, of which 1.9 million were violent incidents (Office for National
Statistics, 2013). Therefore it is vital to identify factors that increase the risk
of violent and aggressive behaviour. A large amount of research has been
conducted to explore the causes and moderators of aggression, focusing on
environmental and individual difference constructs such as aspects of the
self. Disturbances in aspects of the self have been highlighted as important
risk factors in violent and aggressive behaviour (Baumeister et al., 1996;
Donnellan et al., 2005; Kantor & Jasinski, 1998). The role of the self in
social and psychological functioning is well documented (Campbell,
Assanand & Di Paula, 2000), and self-disturbances are prominent features
of many mental health difficulties (Edwards et al., 2012; Fonagy, 1999;
Bateman & Fonagy, 2004), to the extent that they feature in diagnostic
criteria in the DSM-V (APA, 2013). There are a number of theories as to
how an individual’s sense of self may relate to aggressive and violent
behaviour, however these issues have not been comprehensively combined
before, and therefore need to be reviewed.
1.1 Aspects of The Self
235
The self is fundamental to how we experience ourselves, others and the
world around us, and an understanding of the various aspects of the self is
vital to the scientific understanding of personality and behaviour. Decades
of research have detailed the numerous ways in which the self influences
how we think, feel and behave (John, Robins & Pervin, 2010). It is made up
of many facets, including self-esteem, self-concept, identity and self-
efficacy, which will firstly be defined, before outlining the theoretical
explanations linking disturbances in each of these aspects of self to
aggressive behaviour.
1.1.1 Self-Esteem
Self-esteem is an evaluation of worth; the extent to which one sees oneself
as competent and worthwhile (Leary & Tangney, 2005). Until recently self-
esteem has been understood in an uni-lateral way; as either a positive or
negative attitude towards oneself (Gregg & Sedikides, 2003). Now
researchers are starting to understand the full complexity of this multi-
dimensional construct, recognising that conscious and unconscious
evaluations of self-worth constitute self-esteem. Explicit self-esteem is a
conscious, reflective evaluation of the self, whereas implicit self-esteem is
an automatic, reflexive judgement of self-worth. According to The Full
Discrepancy Model (Kernis, 2005), people can have different combinations
of explicit and implicit self-esteem, and a discrepancy between them
represents fragile/insecure self-esteem (Bosson, Brown, Zeigler-Hill, &
Swann, 2003; Jordan, Spencer, Zanna, Hoshino-Browne, & Correll, 2003;
Kernis, 2003).
236
The security of self-esteem plays an important role in psychological
wellbeing and the ability to cope with adverse events (Kernis, 2003). People
with high, secure self-esteem can draw upon internalised feelings of worth
that can be used to maintain positive feelings when negative events occur
(Steele and Aronson, 1995), whereas those who have implicit low self-
esteem alongside explicit high self-esteem (indicating fragile self-esteem)
are less able to maintain such balance (Malle & Horowitz, 1995). This is
because the overall feelings of self-worth of an individual with fragile self-
esteem is injured if they are confronted by negatively perceived events. If
they do not have the skills to enhance their overall self-esteem following a
threat they may use coercive strategies such as aggression to address
imbalances (Quigley, Corbett, & Tedeschi, 1983). This process is
commonly observed in people with high levels of trait narcissism, who are
thought to have fragile self-esteem, constituted of conscious grandiose sense
of self-worth, but underneath an implicit sense of poor self-worth (Campbell
& Foster, 2007; Miller & Campbell, 2008), making their core sense of self-
worth, or ego, vulnerable to injury. These individuals may attempt to defend
themselves from a perceived threat with aggressive behaviour. These
processes are thought to be common in people diagnosed with personality
disorders (PDs), particularly Narcissistic PD (Zeigler-Hill & Jordan, 2011),
which is considered more extreme and problematic than trait narcissism, yet
involves similar deficits in the self (i.e. fragile self-esteem).
Theories have been developed to make sense of the relationship between
self-esteem and aggression. General Strain Theory (Agnew, 2001) suggests
that offending occurs if one experiences strain or stress, and lacks the
237
coping skills required to manage. Low self-esteem may be linked to less
effective coping strategies, thus intensifying the experience of strain and
increasing the likelihood of aggressive behaviour. Conversely, the outcomes
of other studies suggest that high self-esteem is associated with increased
tendencies to express anger and aggression (Baumeister et al., 1996,
Baumeister et al., 2000). These conflicting findings and ideas regarding the
relationship between self-esteem and aggression led researchers to ponder
whether the relationship may be more complex than originally suspected,
which led to a search for other variables that might mediate the relationship.
This led to the Threatened Egotism Hypothesis which predicts that inflated,
yet unstable self-esteem is associated with violence in the context of a threat
to the self or ego (Bushman & Baumeister, 1998). This hypothesis has been
used to explain aggressive behaviour displayed by people with high levels
of narcissism, who have high, but fragile, self-esteem and are vulnerable to
experiencing ego threats (Baumeister & Leary, 1995; Baumeister, Smart &
Boden, 1996; Baumeister et al., 2000; Bushman & Baumeister, 1998;
Baumeister et al., 2002).
Narcissistic PD is characterised by high explicit but low implicit self-worth,
described in the diagnostic criteria as fragile self-esteem (APA, 2013;
Zeigler-Hill & Jordan 2011), and the Threatened-Ego Hypothesis has been
used to help explain aggressive behaviour in these clinical presentations, as
well as psychopathic individuals who are thought to possess very high levels
of narcissist traits (Hare, et al., 1993). There is a very high prevalence of
people with particular PD diagnoses in forensic populations such as
Borderline and Narcissistic PD (Fazel & Danesh, 2002; Kernberg, 1967;
238
New et al., 2004), supporting the idea that fragile, unstable self-esteem at a
clinical level is related to aggression. Whether this hypothesis might shed
light on the origins of offending and criminal levels of aggression remains
to be explored.
1.1.2 Self-Concept
The self-concept consists of factual descriptions of the way in which a
person views him or herself, comprised of the traits and qualities that they
consider themselves to possess (Leary & Tangney, 2005). The extent to
which these self-perceptions are clear and consistent with each other is
called self-concept clarity (Campbell, 1990). Poor self-concept clarity (i.e.
poorly defined and/or inconsistent self-perceptions) is associated with
negative affect and feelings of inadequacy (Usborne & Taylor, 2010), as
well as poor mental health (Campbell, Assandand & Di Paula, 2003).
A poor overall self-concept is associated with negative thinking patterns and
distressing emotional states, which have been found in perpetrators of
domestic violence, referred to in the literature as “batterers” (Berkowitz,
1990; Ragg, 1999). Some researchers suggest that a way of coping with
high levels of negativity in the self-concept involves splitting off different
parts of the self-concept as completely positive or negative which creates a
compartmentalised self structure (self-compartmentalisation; Showers &
Kling, 1996b). A compartmentalised self-structure can have a detrimental
impact on their certainty of themselves, limiting their self-concept clarity
and making them very sensitive to negative and positive feedback
239
(Campbell, 1990; Riketta & Ziegler, 2007), and more likely to behave
aggressively in response (Stuke & Sporer, 2002).
Self-discrepancy theory (Higgins, 1987) suggests that individuals have three
different conceptions of the self; the representations of the attributes they do
possess (actual self), the ones they should have (ought self), and what the
individual thinks they would ideally like to possess (ideal self). The theory
claims that a discrepancy between two of these self-representations causes
distress, and people are therefore motivated to reduce the gap. There is some
indication that people with a large gap between self-representations,
particularly the ought and ideal self are more likely to experience high
emotional reactivity to events and display aggressive behaviour in an
attempt to address the imbalance (Kinney, Smith & Donzella, 2001). This
pattern has been observed in domestic violence perpetrators in that they tend
to have unrealistically high standards which they are unable to achieve so
they attempt to bridge this gap by controlling and dominating others in order
to reduce feelings of insecurity (Stets, 1995). These discrepancies make it
difficult for individuals to cope with threats and negative events, as their
self-conception will shift depending on the situation (Campbell et al, 1996).
This has been linked to the development of a range of mental health
problems and social difficulties, such as depression, anxiety, violence,
substance abuse and PDs. According to The Diagnostic and Statistical
Manual of Mental Disorders (DSM V; APA, 2013), negative or unstable
aspects of self are key components of the diagnostic criteria for PDs in
particular, including Borderline and Narcissistic PD.
1.1.3 Identity and Identity Diffusion
240
Identities refer to the characteristics, traits, roles and membership of social
groups that define who one is, which leads to one’s awareness of oneself as
a separate and unique individual (Leary & Tangney, 2005). Identity
disturbances are often referred to in the literature as identity diffusion, and
are thought to play a key role in the relationship between PD pathology and
violent behaviour. Identity diffusion refers to the lack of integration between
one’s self-identities, leading to an overall poorly integrated sense of self
(Clarkin, Yeomans, & Kernberg, 2006). Together, identities make up one’s
self-concept, and the extent to which they are well or poorly integrated is
important in PD pathology. Self-concept clarity and identity diffusion both
appear to refer to poorly integrated self-representations, with the former
term predominantly used in social psychology literature while the latter is
mainly found in the psychodynamic literature.
In the psychodynamic literature, Fonagy’s (1999) attachment theory
formulation of violence proposes that aggression results from an inability to
mentalise the impact of behaviour on another person. The ability to
mentalise, or represent states of the minds of the self and others, is thought
to develop in childhood through having internal states understood by
another via a secure attachment (Fonagy et al., 2002). This attachment
enables children to develop a stable, integrated identity, which enhances
their understanding of their own and others’ feelings (Bateman, Fonagy et
al, 2003), enabling them to flexibly adapt to various situations.
Mentalisation allows one to modulate emotional states and regulate them,
which includes the ability to inhibit aggressive impulses when experiencing
distress (Fonagy & Target, 2006). People with PDs, particularly Borderline
241
PD, have poor mentalisation skills (Bateman & Fonagy, 2004), which may
be linked to the diffused identity they commonly experience, as well as a
tendency to exhibit aggressive behaviour. Additionally, people who have a
diagnosis of Narcissistic PD (NPD) are thought to have high levels of
identity diffusion (referred to in social psychology literature as poor self-
concept clarity), whereby their ego, or core sense of self, is sensitive to
threats (Edwards et al, 2012; Ronningstam, 2012). Accordingly, the
Threatened Egotism Hypothesis (Baumeister, Smart, & Boden, 1996)
predicts that the presence of an ego-threat combined with the fragile self-
concept seen in NPD is likely to result in aggressive behaviour.
1.1.4 Self-Efficacy
Self-efficacy is a person's belief in their ability to accomplish specific goals
or tasks, corresponding to the level of competence they feel that they have
(Ormrod, 2006). It is a vital mechanism of agency, influencing how people
act, feel, and think (Bandura, 1992). Individuals who believe they can
control potential threats, are more likely to be able to cope with stressors,
whereas those with poor self-efficacy in their ability to deal with threats
may feel the need to prove their ability be in control through an aggressive
response (Bandura, 1992). Alternative findings from studies of children
indicate that high levels of self-efficacy for performing aggressive acts are
positively related to violent behaviour (Camodeca & Goossens,
2005), suggesting that beliefs about capabilities are
important drivers of behaviour.
1.4 Summary and aims
242
How one sees, understands, and evaluates oneself has a range of
consequences for functioning and wellbeing, and also for behaviour. There
is a growing empirical focus on the relationship between the various aspects
of the self and their relationship with aggressive and violent behaviour. The
purpose of this review is to systematically review the quantitative literature
pertaining to this relationship.
Much of the research in this area has been conducted within the general
population, however, the patterns that have emerged have also been
observed in forensic settings (Edwards & Bond, 2012; Ragg, 1999). Further
research in this area will enhance our understanding of the relationship
between self-esteem and aggression, and could be used to predict recidivism
and future risk of violence. There are compelling reasons to expect that
aspects of the self will impact on criminal levels of aggression because the
evidence shows a strong link between PDs and aggression and deficits in the
self which lie at the heart of PDs. Individuals with PDs, particularly
narcissistic and borderline PDs, are frequently present within prison
populations. Given the diversity of research and theories on aspects of the
self in previous research, a systematic approach is needed to identify aspects
of the self that are the most relevant in these populations.
Methodology
2.1 Search strategy and selection criteria
The literature search was undertaken using two databases; “Ovid” and “Web
of Knowledge”. These searches were conducted independent of each other
on 18th February 2014. In order to capture all of the search strings related to
243
aspects of self, the following search terms were used: “self identi$”, “self
image$”, “self concept$”, “self-esteem”, “self-stabili$”, “splitting”, “self
complex$”, “self aware$”, “self-other”, “self-affirm$”, “self-effic$”, “self
disturb$, “self worth”, and “identity diff$”. The terms “aggress$” and
“violen$” were used to capture studies investigating aggressive and violent
behaviour. In order to encapsulate studies using samples of participants with
PDs or forensic issues the following keywords were used: “clinical$”,
“psychiat$”, “psychol$”, “mental disord$”, “mental health$”, “mental ill$”,
and “mentally ill”. A DSM diagnosis for PD was not the sole inclusion
criteria because some services adopt formulation-based or alternative
approaches to identify people with these difficulties. Whether the sample
included people with forensic issues or PDs was determined during deeper
scrutiny. Only peer-reviewed articles, written in English were included in
the search from Ovid. Only papers from 1990 onwards were included due to
the time and space restraints on this literature review. By including only
papers from this point onwards I expect to capture the most up-to-date
research. This year (1990) was chosen because this was the year in which
widely publicised efforts to bolster research into aspects of self were made
(California Task Force, 1990), and there was little controlled, quantitative,
peer-reviewed research in the area before this time (Mruk, 2006). Due to the
large number of articles from Web of Knowledge, articles that were not
published in peer-reviewed journals were manually excluded. All articles
obtained from both databases were exported into EndNote, and duplicates
were removed. The inclusion and exclusion criteria are detailed in Table 1.
244
Table 1. Inclusion and exclusion criteriaInclusion Exclusion
Papers published during or after 1990
Written in English Reports of original data only Reports of direct physical
aggression including sexual offences
Papers using adult samples recruited either for personality disorders or forensic issues as the primary problem
Papers that include direct measures of self-variables
Papers reporting secondary data (such as meta-analyses or systematic reviews)
Children/adolescents under the age of 18
Non-peer reviewed journals Non-physical violence Samples with physical health
problems as the primary difficulty
1.2 Method of review
A review of the literature was performed following the Prisma guidelines
(Moher et al., 2009). Titles, and then the abstracts of the remaining papers
were reviewed to ensure that they met the inclusion criteria. The remaining
papers were subject to deeper scrutiny, and were removed at this stage if
they were not research papers reporting primary data from direct,
quantitative measures of aspects of self or physical aggression/violence.
Papers that used samples of participants under the age of 18, or samples that
were recruited for reasons other than their offending/forensic or personality
disorder difficulties were also excluded. Papers excluded at each stage of
analysis are presented in a flow chart (see Figure 1).
245
Figure 1. Flow chart outlining stages of literature scrutiny and reasons for exclusion.
246
To assess the quality of the remaining articles, a checklist for assessing the
quality of quantitative studies developed by Kmet, Lee and Cook (2004)
was used for each paper, which uses an operational scoring system, with by-
item agreement ranging from 73% to 100%. Quality scores were calculated
using the quality assessment scoring instructions for quantitative studies
outlined in Kmet et al. (2004).
2. Results
A summary of the 19 studies included in the review is shown in Table 2,
which outlines the self and aggression measure that was used, as well as
information about the sample of participants and the context of each study.
Key aspects of self related to aggression identified in these papers were self-
esteem, self-concept, identity diffusion and self-efficacy. Findings of papers
exploring the each of these will be reported and critically evaluated in this
section.
247
Table 2. Summary of characteristics of the 19 studies included in the systematic reviewStudy Groups and group size Context of PD / offender
groupStudy design Self-concept measure Aggression/
violence measureQuality Score
Baker & Ireland (2007)
60 male offenders (42 violent and 18 non-violent)32 students
Medium-secure prison ANOVA and regressionBetween-groups
Self-esteemTSBI
Membership of violent/non-violent group
0.82
Barnett et al. (2012)
3402 convicted male sex offenders
Completed treatment in the community through a probation service.
Regression analysis Self-Esteem Scale (Thornton et al. 2007)
Reconviction data (type of offence recorded – sexual or violent)
0.9
Barnett et al. (2013)
3402 convicted male sex offenders
Completed a cognitive-behavioural sexual offender treatment program whilst on probation.
Clinically significant and reliable change analyses andregression analysis.
Self-Esteem Scale (Thornton et al. 2007)
Reconviction data (type of offence recorded – sexual or violent)
0.95
Beesley & McGuire (2009)
60 male offenders (30 violent offences and 30 property offences)30 non-offenders from a community sample
Actively serving custodial sentences at a prison in England.
Cross-sectional comparison design
Self-esteemCFSEI-2Self-conceptSQ
Membership of offender-violent, offender-property or control group
0.95
Cale & Lilienfeld (2006)
98 male violent offenders Prison Within-subjects Self-esteemRSES
Informant reports of aggression and direct reports of aggression following ego threats
0.95
Cohen et al. (2002) 20 male sex offenders (paedophiles)24 healthy controls
Outpatient clinic for sex offenders
Between-subjects Self-conceptDAPI-Q (self-organisation cluster)
Member of paedophile or control group
0.77
Fruehwald et al. (1998)
53 male sex offenders Long-term incarcerated prisoners
Correlational Self-conceptFSKNSelf-esteem subscale of FSKN
Number of previous offences and degree of violence
0.7
248
Study Groups and group size Context of PD / offender group
Study design Self-concept measure Aggression/violence measure
Quality Score
Fruehwald et al. (2000)
38 male sex offenders (17 with and 21 without minimal brain abnormalities)
Maximum secure hospital for mentally disordered offenders
Between-subjects, correlational design
Self-conceptFSKN
FAF 0.58
Gillespie (2005) 644 male violent offenders Offenders service prison sentences cross 19 state correctional facilities in the USA
Cross-sectional survey design andbetween-subjects based on race
Self-esteemRSES
Violent behaviour prior to incarceration and current violent
0.8
Hubbard (2006) 280 male and female violent offenders
Community-based correctional facility, therapeutic community
Between-subjects (grouped according to sex and race)
Self-esteemRSES
Recidivism from arrest data
0.8
Koenigsberg et al. (2001)
152 personality disorder patients
Mood and Personality Disorder Program
Correlational design and regression
DSQ – splitting subscaleRatings on each of the DSM-III-R BPD criteria for identity disturbance
BDHI – assaultiveness subscale
0.75
Loffler-Stastka et al. (2003)
20 (10 male and 10 female) psychiatric inpatients with BPD
Treatment with psychoanalytically oriented psychotherapy for 6 weeks
Between-subjects (users and non-users of psychotherapy)
Questionnaire for Competence & Control Convictions – self-concept subscale
Questionnaire for Assessing Aggression Factors
0.75
Long, Fulton, Dolley & Hollin (2011)
44 women with stable psychiatric symptoms (31 PD primary diagnosis and 10 PD as secondary diagnosis) detained under MHA (most with violent index offence)
Two medium secure wards. All participants participating in a therapeutic group
Pre-test post-test design to evaluate a group program
Self-efficacyGSES
Overt Aggression Scale 0.8
249
Study Groups and group size Context of PD / offender group
Study design Self-concept measure Aggression/violence measure
Quality Score
Marshall, Barbaree & Fernandez (1995)
55 sex offenders20 community subjects20 students
Outpatient sexual behaviour community clinic
Between-subjects Self-esteemSSEI
Member of sex-offender, community or student group
0.77
Marshall & Mazzucco (1995)
24 sex offenders (child molesters)23 non-offenders
Bath Institution Sex Offenders Treatment program
Between-subjects Self-esteemSSEI
Member of sex-offender or non-offender sample
0.77
Plutchik, Botsis & Van Praag (1995)
79 psychiatric inpatients Inpatient ward Correlational Self-esteemSES
PFAV 0.75
Ragg (1999) 272 male violent offenders Attending batterer treatment program
Between-subjects andregression analysis
Self-conceptPRI
Violence reported on the demographics questionnaireMember of batterer or non-batterer group
0.95
Russell & Jory (1997)
45 abusive and violent male offenders receiving treatment and 16 students
Outcomes of group intervention programs
Within-subjects – pre- and post-treatment.Between-subjects – treatment and comparison group.
Self-esteemRSES
ABI 0.82
Svindseth et al. (2008)
186 acute psychiatric patients Closed acute psychiatric units
Between-subjects (separated into high and low narcissism)
Self-esteemRSES
A scale predicting aggression and dangerousness (violence from first contact to discharge and from observations and medical records)
0.91
250
Key: TSBI – The Social Self-Esteem: Texas Social Behaviour Inventory, Short Form, CFSEI-2 – Culture-Free Self-Esteem Inventory, SQ – Selves Questionnaire, RSES – Rosenberg Self-Esteem Scale, FSKN – Frankfurt Scales of Self-Concepts, FAF – Inventory for the Assessment of Factors of Aggressiveness, BDHI – Buss-Durkee Hostility Inventory, BPI – Borderline Personality Inventory, IPQ – Inventory of Personality Organisation, MPQ – Multidimensional personality questionnaire, GSES – Generalised self-efficacy scale, RCAP – Response Choice Aggression Paradigm – measures physical aggression via electric shock (of opponent) under lab conditions. 3 scores – mean shock duration, intensity and shock frequency., RPAQ – Reactive-Proactive Aggression Questionnaire, SSEI – Social Self-Esteem Inventory, SES – Self-esteem scale, PFAV – Past Feelings and Acts of Violence Scale – ask about acts of violence against others, PRI – personal relations index – measures self-concept attributes in batterers (includes relational self-concept scale and unstable self-concept scale), ABI – Abusive behaviour inventory, TCI - Temperament and Character Inventory (includes a measure of self-concept). PD – personality disorder, BPD – Borderline PD., NPD – Narcissistic PD
251
3.1 Aggression and Self-Esteem
Table 3 summarises outcomes for the 13 studies investigating the
relationship between self-esteem and aggression. Eleven studies used
samples of violent offenders and two studies used a sample of psychiatric
inpatients. The definition of self-esteem appeared to be fairly consistent
across studies. Six studies found no relationship between self-esteem and
aggression (Baker & Ireland, 2007; Barnett et al., 2013; Beesley &
McGuire, 2009; Cale & Lilienfeld, 2006; Marshall, Barbaree & Fernandez,
1995; Plutchik, Botsis & Van Praag, 1995). Three studies found a
significant relationship between low self-esteem and aggression (Barnett et
al., 2012; Fruehwald et al., 1998; Marshall & Mazzucco, 1995), and one
study found that self-esteem and perpetration of physical abuse were both
lower after treatment (Russell & Jory, 1997). One study found that high
self-esteem was significantly associated with violence (Gillespie, 2005).
Two studies found a significant relationship between high self-esteem and
violence when combined with another variable i.e. high levels of narcissism
(Svindseth et al., 2008), or race (Hubbard, 2006). The two studies that
investigated the role that race might play in the relationship between self-
esteem and violence had conflicting findings. Gillespie (2005) found that
self-esteem was a significant predictor of violence in people from a white
racial background only; when self-esteem increased for this client group,
violence also increased. Whereas Hubbard (2006) found that as self-esteem
increased, the likelihood of arrest decreased for this group. For African-
Americans, as self-esteem increased, the likelihood of arrest increased.
These two studies used two different measures of aggression, with Gillespie
252
(2005) using measures of prior and current violence, and Hubbard (2006)
using arrest data to assess recidivism, which could be a reason for the
conflicting findings.
253
Table 3. Summary of Studies Investigating the Relationship Between Self-Esteem and Violence (K=13)Paper Self-Esteem
MeasureAnalysis Strength and Significance of Relationship Other factors looked at in
relation to aggression (*significant)
Quality Score
Baker & Ireland (2007)Violent Offenders
Social Self-esteem: Texas Social Behaviour Inventory (TSBI)
ANOVA to test for group differences and regression to determine membership to offender or non-offender group.
No significant difference between self-esteem between the offender and non-offender groupSelf-esteem was not a significant predictor of membership to either group.
Dyslexic traits*Executive functioningImpulsivity
0.82
Barnett et al. (2012)
Sex Offenders
Self-Esteem Scale (SES)
Regression analysis to see whether pre- or post-treatment scores predicted recidivism
Lower post-treatment self-esteem scores were predictive for recidivism.When post-treatment self-esteem score combined with static risk assessment score, was slightly more accurate at predicting relative risk than static risk alone.
Static riskEmpathy (fantasy subscale*)Relapse prevention*Emotional lonelinessAttitudes towards children and sexAssertivenessLocus of controlImpulsivity
0.9
Barnett et al. (2013)
Sex Offenders
SES Clinically significant and reliable changeRegression analysis
38.6% recovered on self-esteem post treatmentNo significant associations between self-esteem change outcomes and reconvictionNo significant associations between change outcome and reconviction for a sexual or violent offence for those who were functional pre-treatment on self-esteemSelf-esteem had poor predictive accuracy for reconviction
EmpathyRelapse prevention*Emotional lonelinessAttitudes towards children and sexAssertivenessLocus of controlImpulsivity
0.95
254
Paper Self-Esteem Measure
Analysis Strength and Significance of Relationship Other factors looked at in relation to aggression (*significant)
Quality Score
Beesley & McGuire (2009)
Violent Offenders
Culture-Free Self-Esteem Inventory
Looked at 3 different types of self-esteem
ANOVA to see if scores on measures differed between groups
No significant differences were found between any of the three groups, for either General self-esteem, Social self-esteem, Personal self-esteem, Total self-esteem or Lie scale.
Sex roleHypermasculinity*Self-concept*
0.95
Cale & Lilienfeld (2006)
Violent Offenders
Rosenberg Self-Esteem Scale (RSES)
Correlations among self-report measures and informant reports and disciplinary reports of aggression
No relationship between self-esteem scores and measures of aggression in response to ego threats.
Psychopathy (self-report)*Psychopathy*State-trait anger expressionPerceived ego threat*Narcissism*
0.95
Fruehwald et al. (1998)
Sex Offenders
Self-esteem subscale of Frankfurt Scales of Self-Concepts (FSKN)
Correlational analysis looking at relationships between criminal history and self-conceptsMultivariate regression analysis between subscales and criminal history.
Self-esteem was negatively correlated with total number of previous convictions and total number of incarcerations.
Various subscales of self-concept measure*
0.7
Gillespie (2005)
Violent Offenders
RSES Chi-square examining frequencies of categorical dataIndependent samples t-test to look at associations of race with various variables.Multiple regression
Higher self-esteem was positively related to current violence.People from a black racial background had a significantly higher mean score for self-esteem than people from a white background.
VictimisationAggressive personalityPrior drug abuseCurrent drug abuseAge*Race
0.8
255
Paper Self-Esteem Measure
Analysis Strength and Significance of Relationship Other factors looked at in relation to aggression (*significant)
Quality Score
Hubbard (2006)
Violent Offenders
RSES T-testsLogistic regression analyses
Self-esteem as a main effect was not related to arrest for the sample of offendersAfrican-American men and women with high self-esteem were significantly more likely to be arrested than white men and women with high self-esteem.
Controlled for criminal history, substance abuse and attitudes towards crimeRace*Gender
0.8
Marshall, Barbaree & Fernandez (1995)
Sex Offenders
Social Self-Esteem Inventory (SSEI)
Correlational and ANOVA No difference between non-offender groups and pooled results from sexual offenders. Rapists had higher self-esteem than the two child molester groups
Social avoidance and distressAssertiveness*
0.77
Marshall & Mazzucco (1995)
Sex Offenders (against children)
SSEI T-test – difference in scores between child molesters and non-offenders
Significant group differences on the SSEI. Child molesters had lower self-esteem than non-offenders.
Parental acceptance-rejectionFamily violenceChildhood sexual abuse*
0.77
Plutchik, Botsis & Van Praag (1995)Amplifiers and attenuators of violence risk in psychiatric inpatients.
SES Correlational Ego-strength and self-esteem are neutral in relation to violence risk, and they had no correlation with violence risk.
Suicide risk*Depression*Ego strengthPoor reality testing*Sexual driveSexual inhibitionImpulsivity*Sex
0.75
256
Paper Self-Esteem Measure
Analysis Strength and Significance of Relationship Other factors looked at in relation to aggression (*significant)
Quality Score
Russell & Jory (1997)
Violent Offenders
RSES T-tests pre and post-group treatment
Physical abuse was significantly lower post-treatment and lower self-esteem post-treatment.
0.82
Svindseth et al. (2008)
Psychiatric inpatients
RSES Independent samples t-tests and ANOVAs
Severe violence and higher self-esteem scores were significantly associated with high narcissism
Narcissism*Psychiatric symptomsAnxiety and depressionFunctioning
0.91
257
All studies used measures of explicit self-esteem only, indicating a
significant gap in the literature regarding knowledge of the relationship
between implicit self-esteem and aggression. The measures of explicit self-
esteem varied, and therefore may have measured different components of
the construct. For example, Beesley and McGuire (2009) used The Culture-
Free Self-Esteem Inventory-II (CFSEI-2), however, the culture-free nature
of the items is not clear, and the manual presents no evidence that the test is
free of cultural bias (Battle, 1992). The CFSEI-2 reportedly measures
personal, general, and social self-esteem but the definitions of these
dimensions are not described, and it is unclear how each one is separate.
Baker & Ireland (2007) used a measure of social self-esteem (TSBI), which
has been conceptualised as different to global self-esteem (Blascovich &
Tomaka, 1991), although the distinguishing features of each type of self-
esteem is again unclear. The conflicting findings in this area may be
explained by the likelihood that these studies were measuring different
dimensions of self-esteem.
Types of aggression (i.e. direct violent, sexually violent and sexual violence
against children), as well as measures of aggression varied considerably,
and included group membership (i.e. differences between a group of violent
offenders and a non-violent comparison group; Baker & Ireland, 2007;
Beesley & McGuire, 2009; Marshall et al, 1995; Marshall & Mazzucco,
1995; Russell & Jory, 1997), recidivism risk based on previous violence
(Barnett et al, 2012; Plutchik et al, 1995; Svindseth et al, 2008), recidivism
(Barnett et al, 2013; Hubbard, 2007), informant reports (Cale & Lilienfeld,
2006), previous convictions (Fruehwald et al, 1998; Gillespie, 2005), and
258
self-reports (Gillespie, 2005), which could explain the variations in findings.
It is notable that the only study that found a significant relationship between
high self-esteem and higher levels of current violence (Gillespie, 2005) was
also the only study to use a self-report measure of aggression. As this study
was conducted within a prison, inmates may have been motivated to project
an aggressive image and to also report higher levels of self-esteem.
Unfortunately this study did not control for this potential confounding
variable and their findings may have been influenced by socially desirable
responding, rather than reflecting a true relationship between self-esteem
and current aggression.
A broad definition of violence was used in this literature review, therefore
different types of violence measures were included (such as general
violence, sexual, and non-sexual violence). Two studies found a significant
relationship between low self-esteem and violence in a sample of sex
offenders (Fruehwald, et al, 1998; Marshall & Mazzucco, 1995). The latter
study specified that their sample consisted of child molesters, however the
former study did not specify whether the sexual offences were committed
against adults or children. Marshall et al. (1995) found that people convicted
of rape had higher self-esteem than individuals convicted of child
molestation. These findings indicate that lower self-esteem may be related
to sexual offences against children. Two other studies in this review also
used samples of sex offenders (Barnett et al, 2012; Barnett, 2013), however
it was not specified whether the participants were convicted of sexual
offences against adults or children, which may aid the interpretation of the
conflicting findings in this area.
259
Two studies used a sample of psychiatric inpatients which included a sub-
group with a PD diagnosis, although the type of PD was not specified in
either study (Plutchik et al, 1995; Svindseth et al, 2008), and different
diagnostic criteria were used; the DSM-III and ICD-10, respectively.
Svindseth et al (2008) used a large sample and found that high self-esteem
and severe violence on the wards were significantly associated with high
levels of narcissism, whereas Plutchik et al (1995) found no significant
correlation between past acts of violence and self-esteem. However Plutchik
et al (1995) used a considerably smaller number of participants and give no
indication of the appropriateness of the sample size, so it is possible that the
study was under powered. Also they asked participants about past acts of
violence prior to admission approximately 1-2 weeks after admission. One
would expect patients to be acutely unwell on admission to an inpatient
ward so it is possible that they may have found it difficult to remember
previous aggressive behaviour when experiencing distress, and they may
have wanted to present a favourable impression to the staff who will be
working with them during their inpatient stay. Both of these factors may
have influenced participants’ answers on the violence questionnaire,
therefore their answers may not have truly reflected the frequency or
severity of past violent behaviour, which could be why they failed to find a
significant association between self-esteem and aggression.
In terms of quality, all studies achieved a quality score of equal to or greater
than 0.75, apart from Fruehwald et al (1998) due to the unclear description
of the study design, insufficient detail regarding the sample and lack of
justification of the sample size, making it difficult to ascertain the
260
appropriateness of the design and sample, and unclear whether the study
was adequately powered.
Overall eleven studies included a sample of offenders and two used a PD
sample. Findings from PD samples indicate that narcissism may play a role
in the relationship between self-esteem and aggression, while the forensic
literature shows mixed findings, possibly due to the use of a variety of
different measures of aggression, and the various types of convictions of
their participants (sexual, non-sexual and general violence). The conflicting
findings may be explained by the different measures of self-esteem used,
which may be measuring different components of self-esteem. It is
promising that self-esteem is considered alongside other factors such as race
and gender, as well as personality traits such as narcissism when
investigating factors that influence aggressive behaviour. However, the
clinical literature has not yet caught up with non-clinical findings in this
area, and considering self-esteem as a unilateral construct is a prominent
weakness of the existing evidence within forensic and psychiatric
populations because this gives an incomplete picture of the relationship
between self-esteem and aggression. Models of narcissism may help explain
the association between various types of aggressive behaviour and
dimensions of self-esteem.
3.4 Aggression and Self-Concept
Six studies explored the relationship between aggression and self-concept
(see Table 4). Five studies used a sample of violent offenders and one study
used a sample of psychiatric inpatients. A fairly consistent finding was that
261
impaired self-concepts are related to aggression, but the nature of
impairment and impact of a “low self-concept” was unclear (Cohen et al.,
2002; Fruehwald et al, 1998; Fruehwald et al, 2000; Ragg, 1999). One study
made the more specific conclusion that impairments in self-concepts affect
the way that offenders perceived their aggressive behaviour (Fruehwald et
al., 2000), although this study was conducted with people who had “brain
abnormalities”, which is an important individual difference that is likely to
have affected the way their participants perceived their aggression, rather
than “negativity” in their self-concept. Loffler-Stastka et al. (2003) found a
significant relationship between aggression and self-concept in BPD
patients who did not engage in psychotherapy treatment (nonusers).
Although they did not state the direction of the relationship, they suggested
in their discussion that nonusers may have higher self-concept which is
related to an increased likelihood of aggressive behaviour. They linked the
higher self-concept to narcissism, suggesting that nonusers are
narcissistically pre-occupied with the restoration of their selves, and are
therefore less likely to engage in therapy for fear of facing their core,
narcissistic self-concept. They did not, however, make any inferences
regarding the stability of the self-concept and aggression, whereas Ragg
(1999) did, finding a significant positive relationship between an unstable
self-concept and aggression.
Only one study found no significant relationship between self-concept and
aggression (Beesley & McGuire, 2009). It was unclear why the finding in
this study were inconsistent with the others reviewed, as this study had a
good quality score and the sample, design or type of measure used did not
262
appear to differ from those of other studies investigating self-concept and
aggression.
Definitions and measures of self-concept varied considerably in these
studies. Some measures of self-concept also appeared to be measuring self-
efficacy and self-esteem (QAAF used by Loffler-Stastka et al., 2003 and
FSKN used by Fruehwald, 1998 and Fruehwald, 2000). Some studies
referred to self-esteem and self-concept interchangeably in their
methodology and discussion sections (Cohen et al; Fruehwald et al, 1998),
making it unclear what was being measured, or which aspect of self they
were referring to. Loffler-Stastka et al. (2003) claim to be measuring self-
concept yet in their discussion they refer to “self-concept of capabilities”
which seems more like self-efficacy. Although the papers were written in
English, two of them were conducted in other countries (Fruehwald and
Loffler-Stastka) so translation difficulties could be a reason for the
confusion in terminology.
Two studies used a multi-dimensional measure of self-concept (Beesley &
McGuire, 2009; Ragg, 1999), examining at the content of self-concepts as
well as discrepancies between them and stability. The findings from these
studies are more robust, because they consider the multidimensional aspects
of the self-concept, yet the findings were conflicting. A possible reason for
this is that the measure they used (Selves Questionnaire, Higgins et al.,
1985) to measure self-concept does not appear to be well validated, and thus
it may not an appropriate measure of self-concept in offenders. The Personal
Relations Index (PRI) appears to be more robust, with satisfactory reliability
263
and validity reported (Ragg, 1991). These findings therefore constitute
stronger evidence of the association between aggression and self-concept.
As with the self-esteem literature, the types of aggression varied greatly and
included violent offences and sex offences, against adults as well as
children. Of note, all of the studies that used samples of sex offenders found
a significant negative relationship between self-concept and aggression
(Cohen et al, 2002; Fruehwald et al, 1998; Fruehwald et al, 2000),
suggesting that lower self-concepts may be associated with this particular
form of violence.
Measures of aggression also varied; with some studies investigating whether
self-concept was higher or lower in a violent compared to a nonviolent
group (Beesley & McGuire, 2009; Cohen et al, 2002; Ragg, 1999), others
looked at the severity of violence and frequency of previous offences
(Fruehwald et al, 1998), and the other two measures were poorly defined
making it difficult to infer whether they were capturing past or previous
aggression, and it was unclear whether they were self-report or assessment
tools (FAF used by Fruehwald et al, 2000 and Questionnaire for Assessing
Aggressiveness Factors used by Loffler-Stastka et al, 2003). The measures
used did not appear to influence the direction or significance of the
relationship between self-concept and aggression, however, the predominant
pattern was that self-concepts characterised by negativity were associated
with higher levels of aggression. There was a large variation in quality
scores of these papers. The papers which received poorer scores did not
include a justification of the sample size, some did not clearly describe their
sample characteristics or sampling strategy, which may have led to sampling
264
bias, meaning that the findings cannot be generalised to the entire
PD/forensic population. However, the results did not appear to vary
according to quality scores (Fruehwald et al., 1998; Fruehwald et al., 2000).
Overall, the evidence suggests that an impaired self-concept is linked to
aggressive behaviour in both PD and forensic populations. However to what
extent the self-concept is impaired (i.e. whether it is high, low, or unstable)
is unclear. This may be due to the lack of clarity in definitions of self-
concept across the reviewed studies and the variation in measures of self-
concept. The only study to examine various aspects of the self-concept
found that the degree of stability is an important factor.
265
Table 4. Summary of Studies Investigating the Relationship Between Self-Concept and Violence (K=6)Paper Self-Concept Measure Analysis Strength and Significance of Relationship Other factors looked at in
relation to aggression (*significant)
Quality Score
Beesley & McGuire (2009)
Violent Offenders
Selves Questionnaire (SQ) – measures 4 types of self- discrepancies
ANOVA to see if scores on measures differed between groups
No significant differences were found between self-discrepancy/SC scores between control and violent group.
Sex roleHypermasculinity*Self-concept
0.95
Cohen et al. (2002)
Sex Offenders (paedophiles)
Temperament and Character Inventory (includes a measure of self-concept)
Multivariate ANOVA
Compared pedophiles and HCs. Pedophiles have impaired self-concept.
Assertiveness*Passive-aggressiveness*Sociopathy*Cognitive distortions*
0.77
Fruehwald et al. (1998)
Sex Offenders
Frankfurt Scales of Self-Concepts (FSKN)
Correlational analysis looking at relationships between criminal history and self-conceptsMultivariate regression analysis between subscales and criminal history.
Significant correlations between total self-concepts score and total number of incarcerations and total number of previous convictions. Lower self-concept associated with higher number of incarcerations and previous convictions.
Various subscales of self-concept measure*
0.7
Fruehwald et al. (2000)
Sex Offenders
FSKN Correlational analysis
Total self-concept score was negatively correlated with self-perceived aggression only in the MRI negative group – the higher the self-concepts, the lower the self-perceived aggression
Brain abnormalities (MRI)* 0.58
Loffler-Stastka et al. (2003)Psychiatric patients with BPD
Questionnaire for Competence & Control (QCC) Convictions
Correlational Significant correlations between scores for aggression and scores for self-concept in people not engaging with treatment.
AnxietyAbility to relate to other peopleLocus of control
0.75
266
Paper Self-Concept Measure Analysis Strength and Significance of Relationship Other factors looked at in relation to aggression (*significant)
Quality Score
Ragg (1999)
Violent Offenders
Personal Relations Index – separated into 3 self-concept dimensions – USC, RSC and StC
Logistic regression USC and RSC were significantly related to violence grouping (batterers vs nonbatterers) with significant wald values (effect size).Batterers have higher levels of negativity in their self-concept, and less stability in their self-concept.
Verbal violence in family of origin*
0.95
Note – IIP – Inventory of Interpersonal Problems, USC – Unstable Self-Concept Scale, RSC – Relational Self-Concept Scale, StC – Sensitivity to Criticism Scale.
267
3.5 Aggression and Identity Diffusion
One study looked at identity diffusion using a sample of psychiatric
inpatients (Koenigsberg et al., 2001). This study found a positive correlation
between identity diffusion and aggression, but it was not statistically
significant. However it is important to note that participants in this study
appear to have been receiving treatment at the time of participating, and it is
possible it may have impacted on their sense of self-identity and on their
aggressive behaviour. This could have been why the authors failed to find a
significant relationship.
The quality score of this paper was slightly lower than the scores for papers
focused on other aspects of self, mainly because the study design was not
clearly identifiable, the aggression measure was not clearly defined with no
psychometric properties reported, and no estimate of variance was reported
for the main results. These issues make it difficult to identify whether the
study was free from bias, and whether the aggression measure was valid and
reliable. Identity diffusion was rated using the corresponding section of the
DSM-III-R criteria for Borderline PD, using four scores to indicate whether
this characteristic was absent, probably present, present, or strongly present
which seems an overly simplistic rating method of a rather complex
construct. Although the authors report kappa coefficients ranging from 0.64-
0.91, they do not specify the value for identity diffusion ratings, and it is
possible that the inter-rater agreement for this concept was at the lower end
of the range, given that identity diffusion is not clearly defined in this paper.
The authors may have failed to find a significant relationship between
268
aggression and identity diffusion because the measures were not sensitive
enough to capture the true nature of disturbances in identity.
269
3.6 Aggression and Self-Efficacy
One study examined self-efficacy. Long et al. (2011) examined the
effectiveness of Dialectical Behaviour Therapy (DBT) for women with a
diagnosis of PD, and the paper received a good quality score. Although the
direct relationship between self-efficacy and reductions in risk of violent
behaviours following the therapy were not directly explored, the authors
found an increase in self-efficacy following treatment and significant
reductions in risk behaviours which were monitored and grouped according
to an aggression scale for people classed as “completers” of the treatment
program. The fact that the relationship between self-efficacy and aggression
was explored in the context of a treatment program makes it impossible to
comment on the nature of the relationship at baseline, and whether self-
efficacy is related to aggression or not, in a sample of people with
personality disorders. Nevertheless, this relationship is important because it
suggests that reduction in aggression and increase in self-efficacy is
mediated by engagement in DBT.
Discussion
The aim of this systematic review was to explore and summarise research
investigating the relationships between aspects of self and aggressive
behaviour in samples of people with PDs and forensic histories. Major
themes arising from research into four key aspects of self will be
summarised and discussed in relation to further research and relevant
theory.
4.1 Aggression and Self-Esteem
270
The research from two studies exploring the relationship between self-
esteem and violence in personality disorders was inconclusive (Plutchik, et
al., 1995; Svindseth, et al., 2008). Both of these studies used a sample of
participants on an inpatient ward, who were likely to be receiving treatment,
although this was unclear in both studies. It is important that the effect of
treatment is accounted for in future research, as this could have a significant
impact on an individual’s perception of their self-worth, and could be a
confounding factor. None of the studies reviewed in this paper explored the
full self-esteem construct, so one cannot say whether self-esteem is linked to
aggression in PD or violent populations.
Findings regarding the relationship between self-esteem and aggression in
forensic populations varied considerably. Studies which explored other
factors that might influence this relationship suggest that the concept of self-
esteem might function differently depending on individual and situational
differences such as a person’s racial background, perception of ego-threat,
and levels of narcissism. Therefore, future research should include measures
of these other factors to see if they moderate the relationship between self-
esteem and aggression. These findings have important implications for the
treatment of offenders, suggesting that, rather than delivering treatment
programs that target offenders’ self-esteem, treatment plans should be
developed according to the individual needs of the offenders, and cater for
whether they are actively serving time in prison, or in the community. For
example, following an individual assessment, The Choices, Actions,
Relationships and Emotions (CARE) treatment program aims to teach
offenders how to regulate their emotions by creating a positive self-identity,
271
and various mindfulness and emotional literacy programs aim to improve
self-esteem and self-efficacy in the ability to be responsive to others’
emotions (Blanchette & Brown, 2006; Knight & Modi, 2014). The findings
of the present review suggest that existing treatment approaches should go
beyond increasing self-esteem, focusing instead on developing stability in
one’s perception of self-worth, and increasing implicit, as well as explicit,
self-esteem in order to enhance well-being and reduce the risk of violence.
General Strain Theory (Agnew, 2001), suggests that low self-esteem is
related to violence because people with low self-esteem are thought to have
poor coping skills in relation to stressful events. Therefore the experience of
strain is intensified, increasing the likelihood of aggression and offending.
There was some support for the link between low self-esteem and
aggression (Barnett et al., 2012; Fruehwald, 1998; Marshall & Mazzucco,
1995), although the design of these studies did not allow one to infer
whether poor coping skills were associated with this relationship. Other
studies found the opposite, that it is high self-esteem (Gillespie, 2005), or
that high self-esteem combined with narcissism (Svindseth et al., 2008), or
race (Hubbard, 2006), is associated with increased violent behaviour, which
does not support The General Strain Theory.
The Threatened Egotism Hypothesis predicts that an individual with
unstable or fragile self-esteem is more likely to display aggressive
behaviour in response to an ego-threat (Bushman & Baumeister, 1998). This
hypothesis was not supported by the one study that investigated this
relationship in a sample of offenders (Cale & Lilienfeld, 2006). However,
the study used a global, explicit measure of self-esteem, and did not assess
272
the stability of the self-esteem, a factor that Baumeister claims is an
important mediator of the relationship between self-esteem and aggression.
There are no studies identified in this review that assessed this relationship
in the context of an ego threat, and this is an important research question
required to fully test the Threatened Egotism Hypothesis.
More recent studies are starting to follow the lead of researchers working
with non-clinical populations and are starting to consider the importance of
other variables mediating the relationship between self-esteem and
aggression, such as narcissism (Cale & Lilienfeld, 2006; Svindseth et al.,
2008), which is thought to be strongly associated with unstable and/or
fragile self-esteem (Gregg & Sedikides, 2010), although to date this
research within forensic and personality disordered populations is sparse
with inconsistent findings. Svindseth et al. (2008) found that violence and
self-esteem are significantly associated with high narcissism, whereas Cale
and Lilienfeld (2006) did not find this. However, Cale and Lilienfeld used
informant and disciplinary reports of aggression in prison and their
participants may have been motivated to inhibit aggressive behaviour in this
environment for various reasons. Therefore the environment in which their
study took place may not have been representative of the situations in which
prisoners would normally display aggressive behaviour. Other studies
looking at the relationship between self-esteem and violence found no
significant association.
Overall, findings in this area are inconsistent and conflicting and a key
reason for this could be because they all used self-report measures of
explicit self-esteem only, and thus have not investigated the full
273
dimensionality of the construct. Evidence from the non-clinical literature
indicates that implicit/explicit and stable/unstable aspects of self have
important relationships with aggression (Kernis, 2003; Kernis, 2005;
Edwards and Bond, 2012), so it is important to investigate whether this is
the case in PD and forensic populations. Given the evidence from the non-
clinical literature that self-esteem is a multi-dimensional construct, and that
people can have different combinations of explicit and implicit self-esteem
levels that may be associated with aggression in various ways, future
research should investigate the relationship between this aspect of self and
aggression in a more comprehensive way, which adequately captures the
complexity of the construct.
4.2 Aggression and Self-Concept
A further aspect of the self that has been shown to be relevant to aggressive
behaviour in the present populations is impaired self-concept. The only
study to explore the relationship between self-concept and violence in a PD
sample did so in the context of a psychotherapy treatment program (Loffler-
Stastka et al., 2003). It is possible that the receipt of treatment had an impact
on participants’ self-concepts, and this is an important variable to consider
in future research in this area. Although they did not directly explore the
relationship between these two factors, they did suggest that people with
BPD who did not engage in psychotherapy had a higher self-concept linked
to high levels of narcissistic traits, which was associated with aggression. As
with the literature on self-esteem, it appears that narcissism plays a
moderating role in the association with self-concept and violence in people
with Borderline PD, before they start receiving therapy.
274
Findings from this literature review indicate that, in forensic samples, an
impaired self-concept is consistently related to aggression in sexual offences
in particular (Cohen et al., 2002; Fruehwald et al., 1998, Fruehwald et al.,
2000). Although sexual aggression appeared to be a type of violence linked
to an impaired self-concept, it is unclear how self-concept impairments are
associated with other types of violence, as the measures used by studies
included in this review do not differentiate between various types of
aggression, such as sexual/nonsexual or reactive/instrumental. This is an
important area for further research and may have important implications our
understanding of aggressive behaviour and for the way in which risk is
assessed.
Although most studies found that impaired self-concept is associated with
violence, the extent of impairment of self-concept was unclear, with most
studies assuming that self-concept is a uni-lateral construct (Cohen et al,
2002; Fruehwald et al., 1998; Fruehwald et al., 2000; Loffler-Stastka et al.,
2003; Beesley & McGuire, 2009). Therefore, as with the findings around
self-esteem and aggression, important implicit aspects of self-concept could
have been left unaccounted for, which could have had important
relationships with violent behaviour. The only study to examine various
aspects of the self-concept found that higher levels of negativity and less
stability in the self-concept are important in relation to aggression (Ragg,
1999), which offers some support for The Self-Discrepancy Theory
(Higgins, Klein, & Strauman, 1985). Future research must assess self-
concept in more detail, in order to capture the full complexity of this
construct.
275
It was striking that the research examining self-concepts in relation to
aggression did not clearly define what a self-concept is, and there appeared
to be confusion in many discussion sections regarding what was being
investigated, with self-concept and self-esteem used interchangeably. For
example, Beesley & McGuire (2009) aimed specifically to explore the
impact of self-discrepancy on violent behaviour, yet confusingly, they
discuss their findings from The Selves Questionnaire (designed to measure
discrepancies in self-representation) in relation to self-esteem, which is a
separate, although related, construct. This lack of clarity around the
construct being measured could have been why this study was the only one
to find no significant relationship between self-concept and violence.
Nevertheless, despite the lack of clarity of definitions, there is fairly
consistent evidence that an impaired self-concept is related to aggression,
particularly in violent sexual offences, by people with forensic histories.
4.3 Aggression and Other Aspects of Self
4.3.1 Identity Diffusion
The evidence from the only study in this area indicates that greater levels of
identity diffusion are related to higher levels of aggression in individuals
with a PD (Koenigsberg et al., 2001). Fonagy’s (1999) attachment theory
formulation of violence might help to explain this finding, in that the
inability to mentalise is linked to a fragmented sense of identity, leading to
an increased risk of aggression. However, this study did not include a
measure of mentalisation ability, so future research should include this, to
see whether this ability plays of role in the relationship between identity
276
diffusion and violence. There were no studies investigating identity
diffusion in violent forensic populations.
4.3.2 Self-Efficacy
One paper indirectly explored the association between self-efficacy and
aggression in the context of a treatment effectiveness study (Long et al.,
2011). After receiving group-based therapy based on Dialectical Behaviour
Therapy, self-efficacy improved and aggressive behaviour reduced. Due to
the impact of treatment, one cannot make any conclusions regarding the
relationship between self-efficacy and aggression in people with PD
diagnoses. No studies explored the association between self-efficacy and
violence in forensic populations.
5. Limitations
There are some limitations of this review. This review may be subject to
publication bias, as only quantitative papers published in peer-reviewed
journals were included. Including only quantitative studies may have meant
that qualitative studies providing rich, detailed data on the relationship
between the self and aggression may have been missed. A further limitation
is that only forensic and clinical samples of people who met clinical criteria
were included in this review. Some centres do not use the PD diagnostic
labels and instead consider personality traits as lying on a continuum
(Donnelly, 1998). By focusing on these samples exclusively, potentially
relevant findings regarding aspects of self and violence in clinical
populations may have been missed. The wide definition of aggression might
have confused the findings in relation to self-esteem in particular, as it
277
included generalised, sexual and nonsexual aggression. For example, a
physical assault from a challenged narcissistic person with fragile self-
esteem is quite different from the planned, calculated grooming of a
committed paedophile.
Conclusions and Directions for Future Research
The evidence for the relationship between aspects of self and aggressive
behaviour in violent forensic populations and people with a diagnosis of a
personality disorder is inconsistent.
Future research must explore the full dimensionality of aspects of self in
relation to violence, in order to capture the complexity of the relationship
between aspects of self and aggression in PD and forensic populations.
Although not systematically reviewed in this paper, evidence from non-
clinical populations indicate that the fragility and instability, as well as
implicit aspects of self, are related to increased aggression (Baumeister,
Smart & Boden, 1996; Baumeister et al., 2000), so future research should
investigate these in PD and forensic populations. Identity diffusion is an
important factor in aggressive behaviour in people with a PD diagnosis, and
should be further explored using well validated measures and controlling for
potential variance from treatment. It is important that the samples used
reflect the demographics of offender or personality disorder populations as
much as possible, including a range of racial backgrounds, genders and
ages, given the evidence that these factors may also play a role in the
expression of aggressive behaviour. Further research may inform treatment
278
and/or rehabilitations programs and may help reduce the risk of recidivism
and future violence.
Additionally, there is little research looking at PDs within a forensic context
and this is important because there is overlap between the clinical needs of
these groups in relation to unstable or impaired aspects of self, and both are
at increased risk of displaying violence. Research in this area may have
important implications for developing enhancing the effectiveness of
treatments and for reducing the risk of recidivism.
279
References
Agnew, R. (2001). Strain, personality traits, and delinquency: Extending
general strain theory, Criminology, 40(1), 43-72.
American Psychiatric Association, (2013). Diagnostic and Statistical
Manual of Mental Disorders (5th Ed). Arlington, VA: American
Psychiatric Publishing.
Baker, S. F., & Ireland, J.L. (2007). The link between dyslexic traits,
executive functioning, impulsivity and social self-esteem among an
offender and non-offender sample, International Journal of Law and
Psychiatry 30(6), 492-503.
Barnett, G. D., et al. (2013). Does change in psychometric test scores tell us
anything about risk of reconviction in sexual offenders, Psychology
Crime & Law 19(1), 85-110.
Barnett, G. D., et al. (2012). How useful are psychometric scores in
predicting recidivism for treated sex offenders, International Journal
of Offender Therapy and Comparative Criminology 56(3), 420-446.
Bateman, A.W., & Fonagy, P. (2004). Mentalization-based treatment of
BPD, Journal of Personality Disorders, 18, 36-51.
Battle, J. (1992). Culture-Free Self-Esteem Inventories (2nd Ed). Austin, TX,
PRO-ED.
Beesley, F. & McGuire, J. (2009). Gender-role identity and
hypermasculinity in violent offending, Psychology, Crime & Law
15(2-3), 251-268.
280
Baumeister, R.F., & Leary, M.R. (1995). The need to belong: Desire for
interpersonal attachments as a fundamental human motivation,
Psychological Bulletin, 117, 497-529.
Baumeister, R. F., Bushman, B. J., & Campbell, W. (2000). Self-esteem,
narcissism, and aggression: Does violence result from low self-
esteem or from threatened egotism? Current Directions in
Psychological Science 9(1), 26-29.
Baumeister, R. F., Smart, L. & Boden, J.M. (1996). Relation of threatened
egotism to violence and aggression: The dark side of high self-
esteem, Psychological Review 103(1), 5-33.
Baumeister, R. F., Catanese, K. R. & Wallace, H. M. (2002). Conquest by
force: A narcissistic reactance theory of rape and sexual coercion,
Review of General Psychology 6(1), 92-135.
Berkowitz, L. (1990). On the formation and regulation of anger and
aggression, American Psychologist, 45, 494-503.
Blackburn, R. (2007) Personality disorder and antisocial deviance:
comments on the debate on the structure of the Psychopathy
Checklist – revised, Journal of Personality Disorders, 21, 142– 159.
Blanchette, K., & Brown, S.L. (2006). The Assessment and Treatment of
Women Offenders: An Integrative Perspective, UK: Wiley-
Blackwell.
Blascovich, J., & Tomaka, J. (1991). The Self-Esteem Scale. In J. P.
Robinson, P. R. Shaver, & L. S.Wrightsman (Eds.), Measures of
personality and social psychological attitudes (pp. 115-160). New
York: Academic Press.
281
Boden, J. M., Fergusson, D.M., & Horwood, L. (2007). Self-esteem and
violence: Testing links between adolescent self-esteem and later
hostility and violent behaviour, Social Psychiatry and Psychiatric
Epidemiology 42(11), 881-891.
Bosson, J. K., Brown, R. P., Zeigler-Hill, V., & Swann, W. B., Jr. (2003).
Self-enhancement tendencies among people with high explicit self-
esteem: The moderating role of implicit self-esteem, Self and
Identity, 2, 169-187.
Bushman, B. J. & R. F. Baumeister (1998). Threatened egotism, narcissism,
self-esteem, and direct and displaced aggression: Does self-love or
self-hate lead to violence?, Journal of Personality and Social
Psychology 75(1), 219-229.
Cale, E. M. & Lilienfeld, S.O. (2006). Psychopathy factors and risk for
aggressive behavior: A test of the "threatened egotism" hypothesis,
Law and Human Behavior 30(1), 51-74.
California Task Force to Promote Self-esteem and Personal and Social
Responsibility. (1990). Toward a State of Self-Esteem. Sacramento:
California State Department of Education.
Campbell, J.D. (1990). Self-esteem and the clarity of the self-concept,
Journal of Personality and Social Psychology, 59(3), 538-549.
Campbell, J.D., Assand, S., & Di Paula, A. (2000). Structural features of the
self-concept and adjustment, In A. Tesser, R.B. Felson, & J.M. Suls,
(Eds). Psychological Perspectives on Self and Identity (pp 115-138).
Washington, DC, US: American Psychological Association.
282
Campbell, K.W., & Foster, J.D. (2007). The narcissistic self: Background,
an extended agency model, and ongoing controversies. In C.
Sedikides & S.J. Spencer (Eds) Frontiers in Social Psychology: The
Self (pp 115-138). UK: Psychology Press.
Clarkin, J.F., Yeomans, F.E., & Kernberg, O.F. (2006). Psychotherapy of
Borderline Personality: Focusing on Object Relations. Washington,
D.C.: American Psychiatric Publishing, Inc.
Cohen, L. J., et al. (2002). Personality impairment in male pedophiles,
Journal of Clinical Psychiatry 63(10): 912-919.
Donnellan M.B., Trzesniewski, K.H., Robins R.W., Moffitt, T.E., & Caspi,
A. (2005). Low self-esteem is related to aggression, antisocial
behavior, and delinquency, Psychol Sci, (4), 328-35.
Donnelly, J.P. (1998). A continuum model of personality disorder,
Psychological Reports, 83(2), 387-91.
Edwards, R. & A. J. Bond (2012). Narcissism, self-concept clarity and
aggressive cognitive bias amongst mentally disordered offenders,
Journal of Forensic Psychiatry & Psychology 23(5-6), 620-634.
Fazel, S., & Danesh, J. (2002). Serious mental disorder in 23000 prisoners:
a systematic review of 62 surveys, Lancet, 359(9306), 545-50.
Fonagy, P. (1999). Male perpetrators of violence against women: An
attachment theory perspective, Journal of Applied Psychoanalytic
Studies, 1, 7-27.
283
Fonagy, P.,Gergely,G., Jurist, E., et al. (2002). Affect Regulation,
Mentalization and the Development of the Self. New York: Other
Press.
Fruehwald, S., et al. (2000). Self-concepts and interpersonal perceptions of
sexual offenders in relation to brain abnormalities, Journal of
Psychology & Human Sexuality 11(3), 49-56.
Fruehwald, S., et al. (1998). The relevance of self-concepts discriminating
in long-term incarcerated sex offenders, Journal of Behavior
Therapy and Experimental Psychiatry 29(4), 267-278.
Gillespie, W. (2005). Racial differences in violence and self-esteem among
prison inmates, American Journal of Criminal Justice 29(2), 161-
185.
Gondolf, E.W. (1995). Alcohol abuse, wife assault, and power needs, Social
Service Review, 69(2), 274-284.
Gregg, A.P., & Sedikides, C. (2010). Narcissistic fragility: Rethinking its
links to explicit and implicit self-esteem, Self and Identity, 9, 142-
161.
Hare, R. D., Hart, S. D., Forth, A. E., Harpur, T. J., & Williamson, S. E.
(1993). Psychopathic personality characteristics: Development of a
criteria set for use in the DSM-IV antisocial personality disorder
field trial, DSM-IV Update, Office of Research, American
Psychological Association, 6-7.
Hare, R. D. (1996). Psychopathy and antisocial personality disorder: A case
of diagnostic confusion, Psychiatric Times, 13(2), 39-40.
284
Hervé, H., Hayes, J. P., & Hare, R. D. (2003). Personality and Individual
Differences, 35(7), 1497-1507.
Higgins, E.T. (1987). Self-discrepancy: A theory relating self and affect,
Psychological Review, 94(3), 319-340.
Higgins, E.T., Klein,R., & Strauman, T. (1985). Self-concept discrepancy
theory: A psychological model for distinguishing among different
aspects of depression and anxiety, Social Cognition, 3(1), 51-76.
Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M.
(2005). A meta-analysis on the correlation between the implicit
association test and explicit self-report measures, Personality &
Social Psychology Bulletin, 31(10), 1369-1385.
Hubbard, D. J. (2006). Should we be targeting self-esteem in treatment for
offenders: Do gender and race matter in whether self-esteem
matters?, Journal of Offender Rehabilitation 44(1), 39-57.
Institute for Economics & Peace (2013). The Economic Cost of Violence
Containment: A Comprehensive Assessment of the Global Cost of
Violence. Institute for Economics & Peace, Australia.
John, O.P., Robins, R.W., & Pervin, L.A. (2010). Handbook of Personality,
Third Edition: Theory and Research. UK: The Guilford Press.
Jordan, C. H., Spencer, S. J., Zanna, M. P., Hoshino-Browne, E., & Correll,
J. (2003). Secure and defensive high self-esteem. Journal of
Personality and Social Psychology, 85, 969–978.
285
Kantor, G.K., & Jasinski, J.L.(1998). Dynamics and risk factors in partner
violence. In: Jasinski JL, Williams LM, (Eds). Partner violence: a
comprehensive review of 20 years of research. Thousand Oaks, CA,
Sage.
Kernberg, O. (1975). Borderline Conditions and Pathological Narcissism.,
New York: Aronson.
Kernis, M. H., Abend, T., Goldman, B. M., Shrira, I., Paradise, A. W., &
Hampton, C. (2003). Discrepancies between explicit and implicit
self-esteem and self-serving responses, Self and Identity, 4, 311 –
330.
Kernis, M. H., Abend, T. A., Goldman, B. M., Shrira, I., Paradise, A. N., &
Hampton, C. (2005). Self-serving responses arising from
discrepancies between explicit and implicit self-esteem, Self and
Identity, 4(4), 311-330.
Kirkpatrick, L.A., Waugh, C.E., Valencia, A., & Webster, G.D. (2002). The
functional domain specificity of self-esteem and the differential
prediction of aggression, Journal of Personality & Social
Psychology, 82(5), 756-767.
Kmet, L.M., Lee, R.C. & Cook, L.S. (2004) Standard Quality
Assessment Criteria for Evaluating Primary Research Papers
From a Variety of Fields. Alberta Heritage Foundation for
Medical Research.
Knight, C., & Modi, P. (2014). The use of emotional literacy in work
with sexual offenders, Probation Journal, Published online
before print March 27, 2014.
286
Koenigsberg, H. W., et al. (2001). Are the interpersonal and identity
disturbances in the borderline personality disorder criteria linked to
the traits of affective instability and impulsivity?, Journal of
Personality Disorders 15(4), 358-370.
Krizan, Z., & Suls, J. (2008). Are implicit and explicit measures of self-
esteem related? A meta-analysis for the Name-Letter Test,
Personality & Individual Differences, 44(2), 521-531.
Leary, M.R., & Tangney, J.P. (2005). The Handbook of Self & Identity, New
York: London: Guilford Press.
Leichsenring, F., et al. (2003). Borderline personality organization in violent
offenders: Correlations of identity diffusion and primitive defense
mechanisms with antisocial features, neuroticism, and interpersonal
problems, Bulletin of the Menninger Clinic 67(4), 314-327.
Lenzenweger, M. F., et al. (2012). Predicting domains and rates of change
in borderline personality disorder, Personality Disorders: Theory,
Research, and Treatment 3(2), 185-195.
Lenzenweger, M. F., et al. (2012). Exploring the interface of
neurobehaviorally linked personality dimensions and personality
organization in borderline personality disorder: The
Multidimensional Personality Questionnaire and Inventory of
Personality Organization, Journal of Personality Disorders 26(6),
902-918.
287
Loffler-Stastka, H., et al. (2003). Predicting psychotherapy utilization for
patients with borderline personality disorder, Psychotherapy
Research 13(2), 255-264.
Long, C. G., et al. (2011). Dealing with Feelings: The Effectiveness of
Cognitive Behavioural Group Treatment for Women in Secure
Settings, Behavioural and Cognitive Psychotherapy 39(2), 243-2
Malle, B.F., & Horowitz, L.M. (1995). The puzzle of negative self-views:
An exploration using the schema concept, Journal of Personality &
Social Psychology, 68(3), 470-484.
Marcia, J.E. (2006). Ego identity and personality disorders, Journal of
Personality Disorders, 20(6), 577–596 .
Marshall, W., et al. (1995). Some aspects of social competence in sexual
offenders, Sexual Abuse: Journal of Research and Treatment 7(2),
113-127.
Marshall, W. L. & A. Mazzucco (1995). Self-esteem and parental
attachments in child molesters, Sexual Abuse: Journal of Research
and Treatment 7(4), 279-285.
Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., The PRISMA Group.
(2009). Preferred Reporting Items for Systematic Reviews and Meta-
Analyses: The PRISMA Statement. Open Med, 3(3), 123-130.
Mruk, C.J. (2006). Self-Esteem Theory, Research & Practice: Toward a
Positive Psychology of Self-Esteem (3rd Ed.). N.Y: Springer
Publishing Company.
288
New, A.S. et al. (2004). Low prolactin response to fenfluramine in
impulsive aggression, Journal of Psychiatric Research, 38(3), 223-
230.
Office for National Statistics (2013). Statistical Bulletin: Crime in England
and Wales, Year Ending June 2013.
Ormrod, J. E. (2006). Educational Psychology: Developing Learners (5th
Ed.). Upper Saddle River, N.J.: Pearson/Merrill Prentice Hall.
Plutchik, R., et al. (1995). Psychopathology, self-esteem, sexual and ego
functions as correlates of suicide and violence risk, Archives of
Suicide Research 1(1), 27-38.
Quigley, B.M., Corbett, A.B., & Tedeschi, J.T. (1983). Desired image of
power, alcohol expectancies, and alcohol-related aggression,
Psychology of Addictive Behaviours, 16(4), 318-324.
Ragg, D. (1999). Dimensions of self-concept as predictors of men who
assault their female partners, Journal of Family Violence 14(3), 315-
329.
Riketta, M., & Ziegler, R. (2007). Self-ambivalence and
reactions to success versus failure.
European Journal of Social Psychology, 37, 547-560.
Ronningstam, E. (2012). Alliance Building and Narcissistic Personality
Disorder, Journal of Clinical Psychology, 68(8), 943-953.
289
Russell, R. & M. K. Jory (1997). An evaluation of group intervention
programs for violent and abusive men, Australian and New Zealand
Journal of Family Therapy 18(3), 125-136.
Sandstrom, M.J., & & Jordan, R. (2008). Defensive self-esteem and
aggression in childhood, Journal of Research in Personality, 42(2),
506-514.
Short, L.A., et al., (2012). Detection of propensity for aggression based on
facial structure irrespective of face race, Evolution & Human
Behaviour, 33(2), 121-129.
Showers, C. J., & Kling, K. C. (1996b). Organization of self-knowledge:
Implications for recovery from sad mood. Journal of Personality
and Social Psychology, 70, 578-590.
Steele C, & Aronson J. (1995). Stereotype threat and the intellectual test
performance of African Americans, Journal of Personality and
Social Psychology, 69, 797–811
Stets, J.E. (1995). Modelling Control in Relationships, Journal of Marriage
& The Family, 57(2), 489-501.
Stucke, T. S., & Sporer, S. L. (2002). When a grandiose self-image is
threatened: Narcissism and self-concept clarity as predictors of
negative emotions and aggression toward ego threat. Journal of
Personality, 70, 509–532.
Svindseth, M. F., et al. (2008). Narcissism in patients admitted to
psychiatric acute wards: Its relation to violence, suicidality and other
psychopathology, BMC Psychiatry, 8, 13.
290
Toch, H. (1969). Cops and blacks – warring minorities, Nation, 208(16),
491-493.
Usborne, E., & Taylor, D.M. (2010). The role of cultural identity clarity for
self-concept clarity, self-esteem and subjective well-being,
Personality and Social Psychology Bulletin, 36(7), 883-897.
Wallace, M.T., et al. (2012). Locus of Control as a Contributing Factor in
the Relationship Between Self-Perception and Adolescent
Aggression, Aggressive Behaviour, 38(3), 213-221.
Zeigler-Hill, V., & Jordan, C.H. (2011). Behind the mask: Narcissism and
implicit self-esteem. In Campbell, W.K., & Miller, J.D. (Eds). The
Handbook of Narcissism and Narcissistic Personality Disorder:
Theoretical Approaches, Empirical Findings, and Treatments, UK:
John Wiley & Sons.
291
Literature Review Appendix 1
Author Guidelines for Authors From Criminal Behaviour & Mental
Health
Note: No PDF document could be found outlining the guidelines. Instead I
have copied and pasted the information from the website, which is titled
“Author Guidelines”:
http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1471-2857/
homepage/ForAuthors.html
For additional tools visit Author Resources - an enhanced suite of online tools for Wiley Online Library journal authors, featuring Article Tracking, E-mail Publication Alerts and Customized Research Tools.
Author Guidelines
Submission: Papers should be submitted online. For detailed instructions please go to: http://mc.manuscriptcentral.com/cbmh
Previous users can Check for existing account. New users should Create a new account.
Pre-submission English-language editing
Authors for whom English is a second language may choose to have their manuscript professionally edited before submission to improve the English. A list of independent suppliers of editing services can be found at http://wileyeditingservices.com/en/. All services are paid for and arranged by the author, and use of one of these services does not guarantee acceptance or preference for publication.
Copyright and Permissions
Copyright Transfer Agreement
If your paper is accepted, the author identified as the formal corresponding author for the paper will receive an email prompting them to login into Author Services; where via the Wiley Author Licensing Service (WALS) they will be able to complete the license agreement on behalf of all authors on the paper.
292
For authors signing the copyright transfer agreement
If the OnlineOpen option is not selected the corresponding author will be presented with the copyright transfer agreement (CTA) to sign. The terms and conditions of the CTA can be previewed in the samples associated with the Copyright FAQs below:
CTA Terms and Conditions
For authors choosing OnlineOpen
If the OnlineOpen option is selected the corresponding author will have a choice of the following Creative Commons License Open Access Agreements (OAA):
Creative Commons Attribution License OAA
Creative Commons Attribution Non-Commercial License OAA
Creative Commons Attribution Non-Commercial -NoDerivs License OAA
To preview the terms and conditions of these open access agreements please visit the Copyright FAQs hosted on Wiley Author Services and visit http://www.wileyopenaccess.com/details/content/12f25db4c87/Copyright--License.html.
If you select the OnlineOpen option and your research is funded by The Wellcome Trust and members of the Research Councils UK (RCUK) you will be given the opportunity to publish your article under a CC-BY license supporting you in complying with Wellcome Trust and Research Councils UK requirements. For more information on this policy and the Journal’s compliant self-archiving policy please visit: http://www.wiley.com/go/funderstatement.
Permission grants - if the manuscript contains extracts, including illustrations, from other copyright works (including material from on-line or intranet sources) it is the author's responsibility to obtain written permission from the owners of the publishing rights to reproduce such extracts using the Wiley Permission Request Form.
Submitted manuscripts should not have been previously published and should not be submitted for publication elsewhere while they are under consideration by Wiley. Submitted material will not be returned to the author unless specifically requested.
293
Data based papers and systematic reviews are preferred, both quantitive and qualitative research encouraged, but publication of other forms of review or of individual case reports will be exceptional.
Editorials are commonly but not exclusively invited. Anyone wishing to have material published as an editorial should consult the editors prior to writing.
Book reviews are exclusively invited.
Correspondence is welcomed
Title and Abstract Optimisation Information. As more research is read online, the electronic version of articles becomes ever more important. In a move to improve search engine rankings for individual articles and increase readership and future citations to Criminal Behaviour and Mental Health at the same time please visit Optimizing Your Abstract for Search Engines for guidelines on the preparation of keywords and descriptive titles.
Manuscript style. The language of the journal is (British) English. All submissions must have a title, be printed on one side of A4 paper with numbered pages, be double-line spaced and have a 3cm wide margin all around. Illustrations and tables must be printed on separate sheets, and not incorporated into the text.
The title page must list the full title, short title, names and affiliations of all authors. Give the full address, including e-mail, telephone and fax, of the author, who is to check the proofs.
Include the name(s) of any sponsor(s) of the research contained in the paper, along with grant number(s) .
The article will be sent for peer review without the above identifying details.
Substantive research articles should not exceed 3000 words for the main body of the text and contain no more than 4 tables or figures . A structured abstract should be supplied and must not exceed 200 words. It should be subdivided into the following headings: background, aims/hypotheses, methods, results, conclusions/implications for (clinical) practice.
Brief Reports are particularly welcome. They should not exceed 1000 words for the main body of the text, contain only 1 table and/or figure , references should be limited to 20 and an abstract not exceeding 150 words should be supplied.
Reference style. References should be arranged alphabetically. Where reference is made to more than one work by the same author, published in the same year, identify each citation in the text as follows: (Collins, 1998a), (Collins, 1998b). Where multiple authors are listed in the reference, please cite in the text as ‘Maxwell et al. (1999)’. All references must be complete and accurate. Where possible the DOI for the reference should be included
294
at the end of the reference. Online citations should include date of access. References should be listed in the following style:
Website : The Geriatric Website (1999) http://www.wiley.com/oap/ [ 1 April 1999 ]
Paper in a journal : Thierry, KL, Spence MJ (2004). A real-life event enhances the accuracy of preschoolers' recall. Applied Cognitive Psychology 18: 297-309. DOI: 10.1002/acp.965
Book : Hare RD, Schalling D (1978). Psychopathic Behaviour: Approaches to Research. New York: Wiley.
Chapter in a book : Oucho JO, Gould WT, Smith FK, Brown PL, Jones RH (1993). Internal migration, urbanization and population distribution. In Foote KA, Hill KH, Martin LG (eds) Demographic Change in Sub-Saharan Africa. Washington DC: National Academy Press pp. 255-296.
Personal communications and unpublished data should not be listed in the reference section but should be given in full in the text.
Illustrations . Supply each illustration on a separate sheet, with the lead author's name and the figure number (in Arabic numerals), with the top of the figure indicated, on the reverse. Line artwork must be high-quality laser output (not photocopies). Tints are not acceptable; lettering must be of a reasonable size that would still be clearly legible upon reduction, and consistent within each figure and set of figures. Supply artwork at the intended size for printing.
Further information . Proofs will be sent to the author for checking. This stage is to be used only to correct errors that may have been introduced during the production process. Prompt return of the corrected proofs, preferably within two days of receipt, will minimise the risk of the paper being held over to a later issue. Free access to the final PDF offprint of your article will be available via author services only. Please therefore sign up for author services if you would like to access your article PDF offprint and enjoy the many other benefits the service offers. Further offprints and copies of the journal may be ordered. There is no page charge to authors.
295
Clinical Experience
Working Age Adult Placement
Dates: November 2013 – September 2014
Setting: Community Mental Health Recovery Service
Summary of Experience: In this placement I provided CBT, ACT and DBT
psychological interventions to working age adults experiencing moderate-
severe mental health problems such as psychosis, trauma, anxiety, low
mood and personality disorders in a range of community settings.
Psychodynamic processes were explored in supervision. A range of
assessments were undertaken including neuropsychological assessments
followed by feedback, written reports and making referrals. I gained
experience of running groups such as STEPPS, hearing voices groups, and
psycho-education groups. Other experiences included working
collaboratively with charitable sector organisations such as the National
Autistic Society and Mind as well as conducting a service-related research
project investigating service users’ experiences of self-directed support.
Learning Disabilities Placement
Dates: October 2014 – April 2015
Setting: Community Team for People With Learning Disabilities
Summary of Experience: In this placement I supported the mental wellbeing
of adults with intellectual disabilities and their carers in the community in
care homes, day centres, outpatient services and clients’ homes. I conducted
neuropsychological and functional assessments of behaviour that
296
challenges, and assisted a neurodevelopmental service with autistic
spectrum conditions (ASCs) and ADHD assessments. I provided systemic,
DBT and CBT psychological interventions to individuals with autistic
spectrum conditions, Cerebral Palsy and ADHD experiencing a range of
difficulties including bereavement, anxiety, self-harm and “challenging
behaviour”, as well as supporting families and carers. I also ran a group
using CBT and narrative principles to help individuals cope with anger.
Child and Family Placement
Dates: April 2015 – October 2015
Setting: Child and Adolescent Mental Health Service (Tier 3)
Summary of Experience: This placement gave me the experience of
working with young people and families with a range of presenting
problems, including anxiety, depression, self-harm, adapting to and coping
with changes in physical health, trauma and attachment difficulties.
Individual work was undertaken using an integrative approach, drawing on
systemic, narrative, CBT and mindfulness-based approaches. The role
included conducting neuropsychological assessments and sharing findings
with young people, families and schools as well as providing teaching,
supervision and consultations to other professionals including teachers,
social workers and nurses.
Specialist Military Neuro-Rehabilitation Placement
Dates: October 2015 – March 2016
Setting: Defence Medical Rehabilitation Centre (Inpatient and Outpatient)
297
Summary of Experience: This placement involved working with serving
military personnel supporting their recovery and facilitating adaptation to
the effects of traumatic brain injuries, neurological difficulties and
PTSD/adjustment issues. I conducted numerous detailed neuropsychological
assessments, developed formulations and supported the rehabilitation of
various difficulties including traumatic brain injuries. I also offered
interventions for psychological difficulties such as PTSD, complex trauma,
anxiety and low mood. The role involved working with families as well as
supervising assistant psychologists.
Older Adult Placement
Dates: April 2016 – September 2016
Setting: Inpatient Mental Health Ward and Inpatient Dementia Ward
Summary of Experience: This placement involved supporting older adults
living with dementia in an inpatient setting and their family members using
the person-centred care model. I conducted functional assessments of
“challenging behaviour”, developed positive behaviour support plans, led
team formulation sessions, and provided training and consultation to other
health care professionals. I undertook neuropsychological assessments to aid
diagnoses of dementia and to understand the nature of individuals’ cognitive
difficulties and strengths. I also provided group and individual interventions
to older adults with severe and complex mental health problems in an acute
inpatient setting and in the community using CBT and person-centered care
principles.
298
PSYCHD CLINICAL PROGAMME
TABLE OF ASSESSMENTS COMPLETED DURING TRAINING
Year I Assessments
ASSESSMENT TITLE
WAIS Report Short interpretation report of WAIS-IV data
Practice Case Report A Military Veteran Experiencing Trauma-Related Difficulties
Problem Based Learning: Reflective Account
My Relationship With Change
Major Research Project Literature Review
Aggression and Aspects of the Self in Personality Disorders and Forensic Populations: A Systematic Review of the Literature
Adult – Case Report 1 A Neuropsychological Assessment of a Young Man With Epilepsy and Low Mood
Adult – Case Report 2 A Case Report of a CBT Intervention With a Lady Experiencing Social Anxiety and Psychosis
Major Research Project Proposal
The Relationship Between Self-Esteem Fragility and Aggression in a Clinical Forensic Sample
Year II Assessments
ASSESSMENT TITLE
Service Related Project Are Self-Directed Support Packages Helping to Promote Recovery and Independence for People With Complex and Enduring Mental Health Needs?
Professional Issues Essay “Physical Contact With Clients/Service Users is Never Acceptable”. Discuss This Statement in the Context of Your Practice With Clients Across the Life-Span and Specialties.
Problem Based Learning – Reflective Account
Working With High Levels of Risk and Ethical Dilemmas
People with Learning Disabilities – Case Report 3
A Systemic Case Report of a Couple Experiencing Stress Associated With Caring for Their Son With a Moderate-Severe Learning Disability
Personal and Professional Learning Discussion Groups: Process Account
The Growth and Development of Myself and PPD Group
Child and Family – Oral Presentation of Clinical Activity (Case Report 4)
The Story of Luke Pelvis and OCD
299
Year III Assessments
ASSESSMENT TITLE
Major Research Project Empirical Paper
The Relationship Between Self-Esteem, Narcissism, Psychopathy and Aggression in a High Secure Psychiatric Population
Personal and Professional Learning: Final Reflective Account
On Becoming a Clinical Psychologist: A Retrospective, Developmental, Reflective Account of the Experience of Training
Specialist Placement Military Neuro-Rehabilitation – Case Report 5
A Neuropsychological Assessment of a Male Sergeant With a Head Injury, Pain and Memory Problems
300