Offender Profiling PDF

OFFENDER PROFILING: A REVIEW, CRITIQUE, AND AN INVESTIGATION OF
THE INFLUENCE OF CONTEXT, PERCEPTION, AND MOTIVATIONS ON

SEXUAL OFFENDING
By
Shannon Lynn Vettor
A thesis submitted to the College of Life and Environmental Sciences of the University of
Birmingham for the degree of DOCTOR OF PHILOSOPHY
School of Psychology
College of Life and Environmental Sciences
The University of Birmingham
December 2011
University of Birmingham Research Archive
e-theses repository
This unpublished thesis/dissertation is copyright of the author and/or third

parties. The intellectual property rights of the author or third parties in respect
of this work are as defined by The Copyright Designs and Patents Act 1988 or
as modified by any successor legislation.
Any use made of information contained in this thesis/dissertation must be in

accordance with that legislation and must be properly acknowledged. Further
distribution or reproduction in any format is prohibited without the permission
of the copyright holder.
Abstract
This thesis investigates the underlying assumptions of Offender Profiling and the ability to
infer offender characteristics from crime scene characteristics of sexual offences, taking into
consideration the potential mediating effects of the context and situational factors
surrounding the offence, the offender’s perceptions or implicit theories, and their motivations
to sexually offend. Data examined were collected during the evaluation of the Sex Offender
Treatment Programme and consisted of men who committed either a rape or sexual murder
against adult women. Chapter 1 and 2 review the Offender Profiling literature and offer
critiques and areas to further examine, such as the affects of context, perceptions, and
motivations on the A(ctions) to C(haracteristics) equation of Offender Profiling. Chapter 3
outlines the different methodologies used in the subsequent chapters. Chapter 4 compares the
characteristics of rapists and sexual murderers and found very few differences in both
offender and offence characteristics. It also looked at the pathway to offending of sexual
aggressors of adult women and found three, which were supported by previous literature:
Angry, Sadistic, and Sexually Compensatory. Chapter 5 looked at the effects of various
contextual variables on the relationship between offender characteristics and offence
characteristics and found that the location of the offence, and use drugs just prior to the
offence influenced the ability to infer certain offender characteristics from offence
characteristics. Chapter 6 investigates the inclusion of perceptual aspects in the form of the
offender’s implicit theories and the effect these have on the A(ctions) to C(haracteristics)
equation. No significant mediated relationships were found. Chapter 7 examines the effect of
the offender’s motivations to offend and how this would affect the ability to infer offender
characteristics from crime scene variables. Four motivations were found within the current
sample, Angry, Sadistic, Sexually Compensatory, and Sexually Opportunistic. There were no
significant mediating relationships found between the motivations, offender characteristics,
and offence characteristics. The results of the thesis are discussed in terms of both pragmatic
and theoretical support of Offender Profiling and the limitations of the research.
ii
ACKNOWLEDGEMENTS
First and foremost, I’d like to infinitely thank my supervisors Tony Beech and Jessica
Woodhams. While I know it was not always easy, you both provided me with immeasurable
guidance and support, both professionally and personally. Thank you for keeping me going and
giving me the ‘gentle’ nudges I needed to finish.
A special thank you to Sue Hanson for all the time spent listening over the many years
I’ve been at the University, and for always knowing where everything is, and who the correct
person is to speak to! Without you, we would be lost.
To my friends and colleagues in the Centre for Forensic and Criminological
Psychology, who have supported me and been there over the years in many capacities, Leigh
Harkins, Shelley Price, Louise Dixon, and Ross Bartels, without your insights, smiling faces,
constant friendship, and readiness with kind words, the long days and nights would not have
been bearable. For this, I am eternally grateful. To the friends I have met during my time in
England, Eleanor Gittens, Kerri Badger, Katerina Kantartzis, and Kate Graham your ability to
make me cry with laughter and keep me sane will never be forgotten. You have all made my
experience in England more than I could have imagined and hoped for. To Laura, Nattalie,
Lynn, Katie, and Lillian, thanks for all of your constant badgering to come home I miss you
too!
Jay, from the day we met, you have been there; through the good and the bad, even
when you didn’t have to. You have pushed and encouraged me when I couldn’t do it myself. I
am forever grateful for the love and support you and your family have given me.
Finally, I would like to thank my family for their everlasting support and belief in me
over these long years of study, and your constant question: “Are you done yet?!”. To which I
can now answer, “Yes!”. An ocean and half of a country between us has never stopped me
being able to feel your love and support, Mom, Dad, Nana, Papa, and Craig, I thank you, I miss
you greatly, and I love you. To my sister Holly, you have been there since the day I was born,
and I have felt you beside me every step of the way. I love you and I promise not to steer Tyler
wrong!
iii
TABLE OF CONTENTS
Chapter 1: Offender Profiling: An overview and critique of its assumptions 1

and its approaches
1. Introduction 2
1.1 Evaluation of Offender Profiling 3
1.1.1 Utility of Offender Profiles 3
1.1.2 Content of Offender Profiles 3
1.1.3 State of the Literature of Offender Profiling 4
1.2 The Assumptions of Offender Profiling 5
1.2.1 The Behavioural Consistency Assumption 6
1.2.2 The Homology Assumption 7
1.3 Typographical Approaches 9
1.3.1 Pragmatic Criminal Investigative Approach 10
1.3.1.1 Critique of the Early Criminal Investigative Approach 13
1.3.2 Theory-Led Approach 15
1.3.2.1 Critique of the Theory-Led Approach 19
1.3.3 Clinical Approach 20
1.3.3.1 Critique of the Clinical Approach 23
1.3.4 Statistical Approach 24
1.3.4.1 Critique of the Statistical Approach 27
1.3.5 The Current Situation in the United Kingdom: Behavioural Investigative 28
Advice
1.4 Critique of Offender Profiling in General 29
1.4.1 Pragmatic Use and Validity 29
1.4.2 Quality of Data 31
1.5 Summary 32
Chapter 2: Directions for Offender Profiling and Current Thesis 34

2. Introduction 34
2.1 Rationale for Current Research 38
2.2 Context 40
2.3 Perceptions 41
iv
2.4 Motivations 42
2.5 Conclusions 43
Chapter 3: Methodology 46
3.1 Sample 46
3.2 Chapter 4 Methodology 47
3.2.1 Pearson’s Chi-square 48
3.2.2 Independent Sample t-tests 48
3.2.3 Mann Whitney U 49
3.2.4 MANOVA 49
3.2.5 Hierarchical Cluster Analysis 49
3.3 Chapter 5, 6, and 7 Methodology 52
3.3.1 Coding of Variables 52
3.3.1.1 Offender Characteristics 52
3.3.1.2 Offence Characteristics 53
3.3.1.3 Potential Mediators 54
3.3.2 Mediation Analysis 56
3.3.3 Bootstrapping 59
3.3.4 Preacher and Hayes Estimate of the Indirect Effects 60
3.3.5 Assumptions of the Mediation Analysis 61
3.3.5.1 Multicollinearity 61
3.3.5.2 Linear Regression and Logistic Regression Assumptions 62
3.3.6 Measurement of Variance 62
3.3.6.1 R-Square Statistic 62
3.3.6.2 Wald Statistic 63
3.3.6.3 Cox & Snell R and Nagelkerke R Square 63
Chapter 4: Rapists and sexual murderers: Combined pathways to offending 64

4. Introduction 65
4.1 Comparing Rapists and Sexual Murderers 65
4.1.1 Escalation and Continuum of Sexual Violence 67
4.2 Pathways of Sexual Aggression against Women 68
4.3 Methods 71
v
4.3.1 Participants 71
4.3.2 Procedure 72
4.4 Results 73
4.4.1 Findings of the Comparisons between the Rapist and Sexual Murderer 73
Samples
4.4.1.1 Offender and General Life-style Characteristics 73
4.4.1.2 Relationship and Sexual Life-style Characteristics 74
4.4.1.3 Childhood Victimization 74
4.4.1.4 Age at First Offence and Previous Offences 75
4.4.1.5 Pre-crime Factors 75
4.4.1.6 Victim Characteristics 75
4.4.1.7 Modus Operandi 76
4.4.1.8 Mental Health 76
4.4.1.9 Summary of Comparisons 80
4.4.2 Pathways to Offending 80
4.4.2.1 Angry Cluster 82
4.4.2.2 Sadistic Cluster 82
4.4.2.3 Sexually Compensatory Cluster 83
4.5 Discussion 84
4.5.1 Comparison of Rapists and Sexual Murderers 84
4.5.2 Pathways to Offending 85
4.6 Conclusions 87
4.6.1 Suggestions for Future Research 89
Chapter 5, 6, and 7: The Potential Mediators between Crime Scene and 90

Offender Characteristics
Chapter 5: Context as a Mediator of the Interaction between Crime Scene 91
and Offender Characteristics
5. Introduction 91
5.1 Location of Initial Contact and Offence 92
5.2 Victim Resistance 94
5.3 Alcohol and Drug Use 96
5.4 Aim of the Chapter 97
vi
5.5 Method 98
5.5.1 Sample 98
5.5.2 Data Coding 98
5.6 Results 100
5.6.1 Multicollinearity 100
5.6.2 Mediation (Paths a and b), Indirect (ab), and Total Effects 100
5.6.3 Direct Effect of Offence Characteristics (X) on Offender Characteristics 105
(Y)(c’)
5.7 Discussion 106
5.7.1 Limitations 109
Chapter 6: Implicit Theories as a Mediator between Crime Scene and 111

6. Introduction 111
6.1 “Women/Child as Sexual Object” Implicit Theory 111
6.2 “Dangerous World” Implicit Theory 112
6.3 “Male Sex Drive is Uncontrollable” Implicit Theory 112
6.4 “Entitlement” Implicit Theory 113
6.5 “Women are Unknowable/Dangerous” Implicit Theory 113
6.7 Method 116
6.7.1 Sample 116
6.7.2 Content Analysis 116
6.8 Results 119
6.8.1 Occurrence of Implicit Theories 119
(Y)(c’)
6.9 Discussion 124
vii
Chapter 7: Motivations as a Mediator between Crime Scene and Offender 129
Characteristics
7. Introduction 129
7.1 Sexually Motivated: Opportunistic or Compensatory 130
7.2 Anger/Aggressively Motivated 131
7.3 Sadistically Motivated 132
7.5 Method 134
7.5.1 Sample 134
7.5.2 Content Analysis 134
7.6 Results 137
7.6.1 Occurrence of the Motives 137
(Y)(c’)
7.7 Discussion 140
Chapter 8: General Discussion 143

8. Thesis Aims 143
8.1 Summary of Findings 143
8.2 Theoretical and Practical Applications and Limitations 146
8.2.1 Test of the Homology Assumption 147
8.2.1.1 Mediating Variables 147
8.2.1.2 Practical Applications 148
8.3 Limitations 149
8.4 Conclusions 150
References 152
Appendices
Appendix A: Chapter 3 Dendogram Example
viii
Appendix B: Functional Analysis Question
Appendix C: Chapter 4 Comparisons between Rapists and Sexual
Murderers Tables
Appendix D: Figures and Tables of the Context Mediation Analyses
Appendix E: Content Dictionary for Implicit Theories Coding
Appendix F: Figures and Tables of the Implicit Theories Mediation
Analyses
Appendix G: Content Dictionary for Motivations Coding
Appendix H: Figures and Tables of the Motivations Mediation Analyses
ix
LIST OF TABLES
Table Title Page

Chapter 1
1.1 Keppel and Walter’s Classification of Sexual Murder by 18
Motivation, Victim Selection, and Level of Planning
Chapter 3
3.1 List of the Seven Prison Establishments Across England and 47
Wales.
3.2 Offender Characteristics and their Coding 53
3.3 Offence Characteristics and their Coding 54
3.4 Potential Mediators for Chapters 5, 6, and 7, and their Coding 56

Chapter 4
4.1 Range of Base Rate Scores, and Percentage of Rapists and Sexual 78-79
Murderers with Raw Scores Greater than 74 and Greater than 84 on
the Personality and Clinical Syndrome Scales of the Millon Clinical
Multiaxial Inventory-III (MCMI-III)
4.2 Personality, Pre-crime, Sexual Lifestyle and Modus Operandi across 81-82
Offending Pathway
Chapter 5
5.1 Coding of Offence Characteristics (X), Offender Characteristics 99
(Y), and Potential Mediators (M)
5.2 Indirect Effects of Victim Age on Any Previous Convictions 101

through Contextual Variables (ab paths)
5.3 Indirect Effects of a Weapon Being Used on Perpetrator Age 102

through Contextual Variables (ab paths)
x
5.4 Indirect Effects of Sexual Penetration on Perpetrator Age through 102
Contextual Variables (ab paths)
5.5 Total Effects of Offence Variables on Perpetrator Age (c path) 103
5.6 Total Effects of Offence Variables on Relationship Status (c path) 104
5.7 Total Effects of Offence Variables on Any Previous Convictions (c 104

path)
5.8 Total Effects of Offence Variables on Lives Alone (c path) 105
Chapter 6
(Y), and Potential Mediators (Y; Implicit Theories)
6.2 Indirect Effects of Sexual Penetration on Lives Alone through 121

Implicit Theories (ab paths)
6.5 Total Effects of Offence Variables on Previous Convictions (c path) 123

Chapter 7
(Y), and Potential Mediators (Y; Motivations)
7.2 Indirect Effects of Sexual Penetration on Previous Convictions 138

through Motivations (ab paths)
xi
7.5 Total Effects of Offence Variables on Previous Convictions (c path) 140
xii
LIST OF FIGURES
Figure Title Page

Chapter 1
1.1 Ten Step Procedural Model 22
Chapter 2
2.1 Revised A  C: Process of Offender Profiling incorporating 39
Perceptions, Motivations, and Context
2.2 An Aetiological Model of Risk (adapted from Beech & Ward, 2004) 44
Chapter 3
3.1 Simple Mediation Model 58
3.2 Total Effect of X on Y 58

Chapter 5
5.1 Predicting Any Previous Convictions (Y) from Victim Age (X) 100
through the Location of Offence (M).
xiii
Chapter 1 & 2
Offender Profiling: An Overview and Critique of its Assumptions and its Approaches
and the Directions for Offender Profiling and Current Thesis
The aim Chapters 1 and 2 are to provide an overview and critique of the Offender Profiling
literature, its underlying assumptions, the relationship of the A(ctions) to C(haracteristics)
equation at the heart of Offender Profiling, and potential influencing factors on this
relationship. As well, as the various approaches developed over its short empirical history. In
particular, the early Investigative and later Theory-led approaches made public by the FBI,
the Clinical approach, as well as, the more ‘recent’ Statistical approaches will be discussed.
The strengths and weaknesses associated with each of the approaches are evaluated, along
with the potential influences of the offence context, such as location of the offence, the
underlying offender perceptions, their implicit theories, and the offender’s motivations for
offending as prospective directions for Offender Profiling and the rational for the current
thesis.
1
Chapter 1: Offender Profiling: An Overview and Critique of its Assumptions and its
Approaches.
1. Introduction
Offender profiling, while not a new phenomenon, only started developing as an area of
research interest when the FBI Academy in Quantico started publicising their techniques in
the 1970’s (e.g., Douglas, Ressler, Burgess, & Hartman, 1986), which resulted in offender
profiling becoming more widely known (Canter, 2004). Currently, in the UK, the term
offender profiling, or the process by which inferences about potential offender characteristics
are made based on crime scene information, is one aspect of what is referred to as
behavioural investigative advice (BIA; ACPO, 2006). This relates to the analysis of crime
scene behaviours in order to generate both offender characteristics and investigative
suggestions to aid in the generation, prioritisation, and identification of relevant suspects
(Rainbow, 2007).
Generally, offender profiling is referred to as a “technique for identifying the major
personality and behavioural characteristics of an individual based upon an analysis of the
crimes he or she has committed” (Douglas et al., 1986, p. 405). Reiser (1982) has been
attributed as describing profiling as “an arcane art, in which psycho-diagnostic assessment
and psychobiography are combined with case evidence and probabilities from similar cases to
draw a picture of a likely offender” (p. 261). Both are methods that provide possible
descriptions of possible perpetrators of crime based on the analysis of the offence, the
manner in which the offence is committed, and by the determination of personality aspects of
the offender and characteristics from their crime scene actions before, during and after the
offence (Blau, 1994). The underlying rationale is that behaviour will reflect personality, and
by examining behaviour exhibited during the perpetration of a crime an investigator can
determine the likely characteristics of the person responsible for the offence (Douglas et al.,
1986). Simply put, offender profiling is the ability to infer the characteristics of an unknown
offender from their crime scene. The inferences developed are made with the intent to aid law
enforcement personnel in their investigations (Holmes & De Burger, 1988).
2
1.1 Evaluations of Offender Profiling
1.1.1 Utility of Offender Profiles

The utility of profiling was assessed by Copson in 1995. He sampled 184 police
investigations from the United Kingdom where profiling had played a role. On analysis, he
found that profiles could be put into one of two categories: predictive or explanative. The
predictive profiles made predictions about the likely characteristics of the offender, whereas,
the explanative profiles, explained the offender’s behaviour to the police investigators.
Overall, the majority of investigators considered the profiles useful in advancing the case and
stated they would seek profiling advice again. This was despite only 14% of the investigators
stating that the profile helped to solve the case. This is similar to the findings of Jackson, van
Koppen and Herbrink (1993) who found their Dutch police officers to be satisfied with the
profiles included in the study with only two out of 42 evaluations made viewed in a negative
light. In contrast, a nationwide American survey of 152 police psychologists found that 70%
of the police psychologists questioned the validity and usefulness of profiles, and did not feel
comfortable with profiling (Bartol, 1996). Despite concerns about the outcome of using
offender profiles, a contributing factor in their perceived usefulness could be related to the
interpretation of the information contained in them and the potential increase of
understanding of the crime, regardless of the actual pragmatic utility of the profile (Alison,
Smith, & Morgan, 2003).
1.1.2 Content of Offender Profiles

Offender profiling has not only been investigated in terms of its perceived usefulness, in
recent years, two studies have investigated the content of profiles. The first study in 2003, by
Alison, Smith, Eastman, and Rainbow, examined a sample of 21 profiles of violent crimes
(e.g., murder, serial rape, arson), obtained from the United Kingdom, the United States, and
Europe, containing just over 3000 statements. Of the 880 statements referring to the offender
characteristics, 82% were unsubstantiated with no justification of their validity, with less than
16% including any grounds for the claims made. The majority of the statements given in the
profiles were repetition of case facts and offence details already known to the investigating
officers. Comparing Alison et al.’s sample to their UK sample of 47 National Policing
Improvements Agency (NPIA) behavioural investigative advice reports, Almond, Alison and
Porter (2007) examined over 800 claims contained in the reports, finding that 96% providing
some grounds for the claims, but only 34% of these had any formal support or backing. Of
3
these grounded claims, 70% were verifiable, yet more worrying was the finding that only
43% were actually falsifiable, which is an important principle of profiling. This means less
than half could not be proven correct. However, compared to Alison et al.’s (2003) sample of
non-NPIA expert reports, where 82% of their 880 claims were unsubstantiated, and less than
16% included any grounds for the claim, the NPIA expert advice provides more in terms of
substantiation of their arguments with formal grounds for their claims (Almond et al., 2007).
Despite this improvement in the content of the reports, these studies still raise questions
surrounding the proportions of claims which remain without psychological backing (Almond
et al., 2007). However, to be able to substantiate one’s claims in an offender profile requires
there to be sufficient psychological research to be used for this purpose.
1.1.3 State of the Literature of Offender Profiling

Another area in which offender profiling has been subject to critical evaluations is in
terms of the body of literature on this topic. Reviews have concluded that commonsense
rationales are abundant, especially when considering articles written pre-1990, those with a
more clinical orientation, or those authored by law enforcement professionals (Snook,
Eastwood, Gendreau, Goggin, & Cullen, 2007). Typologies of offenders have been a major
product coming out of the profiling literature. The development of a typology, or
classification, of categories of offenders within specific crime types allows, in theory, for the
assignment of a crime scene to a specific category that will have a corresponding category of
offender that exhibits those specific categories of crime scene characteristics with known
characteristics. The premise behind typologies and basic profiling is that each offender type
within the typology is defined by the occurrence of characteristics, behaviours, and crime-
related variables that are representative of it. Within each type the same characteristics co-
occur regularly with each other, and they do not co-occur with the specific characteristics of
any other type of offender (Sarangi & Youngs, 2006). The premise being that each offender
within a type will share the same, or similar, characteristics, behaviours, and crime-related
variables.
Dowden, Bennell, and Bloomfield’s (2007) systematic review of the offender profiling
literature identified at least 132 published articles related to offender profiling since the mid
1970s. What they found was a rapid increase in the amount of offender profiling research, as
well as an increase in the statistical sophistication (e.g., none, descriptive, or inferential) of
the research. It was also found that there was an increase in the amount of studies being
published in peer-reviewed journals, although the majority of these published articles were
4
discussion pieces, basic assumptions, literature reviews, or experiential studies, all of which
provided little or no statistical analyses. Despite this observation, Dowden et al.’s review
shows there’s been an advancement of academic peer-reviewed offender profiling research
(both in volume and statistical sophistication). Even with the advancements in the offender
profiling research, there is still a need for further empirical testing of the assumptions
underpinning profiling (Snook et al., 2007).
1.2 The Assumptions of Offender Profiling
There are two assumptions that must be met if inferences about an offender’s characteristics
are to be derived from crime scene actions, the first is the Behavioural Consistency
assumption which implies that an offender will show similar behaviours across their offences
(Canter, 1995a; Green, Booth, & Biderman, 1976). The second is the Homology assumption,
which states that if two perpetrators exhibit similar crime scene behaviour they will also
possess similar characteristics (Alison, Bennell, Mokros, & Ormerod, 2002; Mokros &
Alison, 2002). The concept of homology was borrowed and adapted from comparative
biology, and was originally defined by Richard Owen (Owen, 2007). The term, defined as
“the same organ in different animals under every variety of form and function” (Owen, 1843;
p. 379), is based on structure, not function. It refers to structures in different species that have
a common evolutionary ancestry; even if they no longer serve the same purpose or have the
same appearance (Owen, 2007). Its adapted use in offender profiling lies in the concept of
similarity between crime scene actions (loosely analogous to structure) and the characteristics
that give rise to the actions (loosely analogous to ancestry). Like the arms of a human, the
foreleg of a dog, the wing of a bird, and the fin of a fish can all be traced back to a specific
original limb in a prehistoric vertebrate, the homologous behaviours exhibited in a crime can
be linked back to a similar characteristic possessed across offenders. The mapping of the
biological sense of homology and the use of the term within offender profiling is not
completely harmonious. However, it is unclear whether the use of the term [by Alison et al.
2002] was meant to be a direct translation to offender profiling, or be used as a distinct
(although tenuously similar) concept specific to offender profiling.
These assumptions are used to validate the “profiling equation”, the A (ctions) to C
(haracteristics) equation (Canter, 2011). The basis for the inferences made in offender
profiling are derived from the actions/aspects of the crime, which ideally would be
behaviourally consistent across an offender’s crimes. These then provide information about
5
the potential characteristics of the offender, with offenders exhibiting similar crime actions
possessing similar characteristic (Canter, 2004; 2011). If these assumptions are invalid, the
practice of offender profiling not only becomes inaccurate, but the advice given to a criminal
investigation would be misleading, resulting in both human and financial costs (Alison et al.,
2002; Gudjonsson & Copson, 1997).
1.2.1 The Behavioural Consistency Assumption

With regards to behavioural consistency, the body of supporting research comes from the
area of case linkage (Bennell & Jones, 2005; Woodhams & Grant, 2006), comparative case
analysis (Bennell & Canter, 2002), or linkage analysis (Hazelwood & Warren, 2003). These
different terms refer to a form of behavioural analysis, based on the behavioural similarity
and distinctiveness of the offence, which is used to identify a series of crimes committed by
the same offender (Woodhams, Hollin, & Bull, 2007), often when there is a lack of physical
evidence (e.g., DNA) to identify the offender (Davies, 1991). Their research supports the
assumption that offenders behave consistently within crime types, including sexual assault
(e.g., Grubin, Kelly, & Brunsdon, 2001; Knight, Warren, Reboussin, & Soley, 1998),
homicide (e.g., Salfati & Bateman, 2005; Santtila et al., 2008), burglary (e.g., Bennell &
Canter, 2002; Bennell & Jones, 2005; Goodwill & Alison, 2007; Green et al., 1976), robbery
(e.g., Woodhams & Toye, 2007), and arson (e.g., Fritzon, Canter, & Wilton, 2001; Santtila,
Fritzon, & Tamelander, 2005).
Common features exist between linkage analysis and offender profiling and some
consider linkage analysis to be another type of offender profiling. The main similarity is that
both share the underlying assumption of offender behavioural consistency; the hypothesis
that offenders will display in their behaviour consistency across a series of offences (Canter,
1995b; Woodhams, Bull, & Hollin, 2007). The possibility of linking a series of crimes can be
beneficial and helpful for the police as it allows for the collection of information across
different crimes, the potential of increasing evidence against an offender, the combining of
separate investigations utilising more effective police resources and lastly, the production of
similar fact evidence in legal proceedings (Grubin et al., 2001; Hazelwood & Warren, 2003;
Labuschagne, 2006). There is growing potential for the utility of linkage analysis in
investigating less serious crimes (e.g., robbery), beyond its current success with investigating
stranger rape or murder, or other more serious crimes (Woodhams et al., 2007).
6
1.2.2 The Homology Assumption
While, studies have shown behavioural consistency of offenders across crimes for different
crime types (e.g., Bennell & Jones, 2005; Grubin, Kelly, & Brunsdon, 2001; Woodhams et
al., 2007), there has been little conclusive support for the assumption of homology. A few
studies have examined the relationship between background characteristics and offence
behaviour and to some extent, support the homology assumption in that bivariate
relationships have been found (Canter & Fritzon, 1998; Davies, Wittebrood, & Jackson,
1998; House, 1997). In their study of 210 solved stranger rapists, Davies et al. (1998)
identified that a rapist’s criminal background could be predicted from his offence behaviour.
Taking precautions not to leave fingerprints (e.g., wearing gloves) was linked with having a
prior custodial sentence, most likely for burglary. As well, an offender who took this
precaution was more likely to be a repeat or serial offender, whereas, when no such
precaution was taken they were three times more likely to be a one-off sexual offender.
Semen destruction indicated a fourfold increase in the offender having been convicted for
prior sexual offences. Theft from the victim and forced entry, were other indicators of prior
convictions for burglary. Making references to the police during the offence and the use of
extreme violence against the victim were an indication of prior convictions for violence.
Therefore, Davies et al. concluded the most promising models were those that predicted
whether the offender had prior convictions for burglary, prior convictions for violent offences
and whether the offence was more likely to be a one-off occurrence rather than committed by
a serial sexual offender.
While, Davies et al. demonstrated that some specific crime scene actions were linked
to particular offender characteristics, their analysis has been criticized for being little more
than predicting associated base rates (Mokros & Alison, 2002). House (1997) investigated
the inverse of the homology assumption – that rapists with different crime scene behaviours
should have different criminal histories. He categorised the rapists as criminal, intimate,
aggressive, or sadistic. In direct opposition to the homology assumption, he instead found a
high degree of similarity in criminality across the four types. Canter and Fritzon (1998)
results looking at the crime scene behaviours from 175 UK arsons were mixed. For three of
the four thematic classifications of arson (instrumental-person, expressive-object, and
expressive-person) there were no conclusive differences between the background
characteristics. The instrumental-object arsons and the expressive-object arsons were the only
two to have positive relations with offender characteristics, the former with a younger
7
offender, and the latter with the individual being a serial arsonist. Doan and Snook (2008)
had similar limited results using a sample of Canadian arsonists who were categorised
according to Canter and Fritzon’s themes. Approximately 56% of the comparisons between
background characteristics and the different thematic classifications of arson violated the
homology assumption. Although, they suggest part of the discouraging findings could be due
to sample differences (UK versus Canadian), and the lack of verification of the arson themes
proposed by Canter and Fritzon. In the same study, Doan and Snook also looked at a sample
of robberies classifying them as Cowboys, Bandits, and Robin’s Men, in accordance with
Alison, Rockett, Deprez, and Watts (2000). The majority of the robberies were classified as
Cowboys or Bandits, with the comparisons between the background characteristics of these
two groups yielding little support for the homology assumption, with 67% of the comparisons
violating this assumption. Woodhams and Toye (2007) also found no support for the
homology assumption in their study of commercial robberies. They examined whether the
commercial robbers could be cluster according to characteristics of the offences, and whether
the background information of these groupings would significantly differ. The resulting three
cluster solution, while providing information about different offending styles, did not differ
significantly according to offender age, ethnicity, employment, previous convictions, or
distance travelled from home to offence.
In Mokros and Alison’s study (2002),the crime scene behaviour and socio-
demographic characteristics and criminal histories of a sample of 100 British male stranger
rapists were examined. Using correlational analysis they tested for similarity between
offenders’ behaviour and characteristics but found no evidence to support the homology
assumption. This indicates that the process of drawing inferences about background
characteristics from crime scene actions is not a simple if A then C equation. Mokros and
Alison offer a possible explanation: the homology assumption, as a simple behaviour-to-
characteristics model, fails because it is too simplistic, and neglects the moderating influence
of a third factor, the situation.
In support of this assumption, Goodwill and Alison (2007) have shown that the
incorporation of the context, such as the level of planning or aggression used, allows for the
prediction of rapists’ characteristics, such as offender age, from their crime scene
information, namely victim age. Alison et al. (2002) suggest that without further
acknowledgement of the situational influence, a direct link between offender characteristics
and crime scene actions is unlikely to be beneficial and until this happens any advice to
police investigations should only be with regards to the prioritisation of suspects as this relies
8
more on behavioural consistency. Although, without the consideration, or understanding, of
how the context or situation influences the perpetration of an offence, the degree of certainty
with which an inference can be made about possible suspects, solely based on behavioural
consistency, needs to be done with caution.
In a novel study, Tonkin, Bond, and Woodhams (2009) found support for the
behavioural consistency assumption, but more importantly, for the homology assumption in a
sample of domestic burglaries. Their results showed the deprivation of the offender’s
residence, and employment status was associated with the expensiveness of their footwear
worn during the crime – the greater footwear costs the greater the association with the
offenders’ residence deprivation and unemployment of the offender. Age and gender did not
relate reliably to the price of footwear worn by the offender. In pragmatic terms, Tonkin et
al.’s results suggest that expensive footwear impressions are related to unemployed offenders
who live in areas of deprivation.
The assumption of behavioural consistency is not dependent upon the assumption of
homology being met or of it being valid; as consistent behaviour across actions does not
necessitate the sameness or similarity of characteristics across offenders (Alison et al., 2002).
That said, behavioural consistency is necessary for offender profiling to work, the offender’s
actions have to remain consistent for similarities to be found between their personal
characteristics and behaviour (Mokros & Alison, 2002). However, the assumption of
behavioural consistency would be valid if the assumption of homology is found to be valid,
due to the implication that similar characteristics imply a consistency of behaviours across
actions (Mokros & Alison, 2002). Yet, for offender profiling to be considered a legitimate,
and importantly, a useful form of behavioural analysis, both of these assumptions must be
met and found to be valid (Alison et al., 2002), as they underpin all forms of profiling.
1.3 Typographical Approaches
The various approaches to profiling are commonly grouped into three schools of thought: a
criminal investigative approach, a clinical practitioner approach, and a scientific statistical
approach (Alison et al., 2010; Muller, 2000). For the current review, the criminal
investigative approach will be further partitioned into the pragmatic and theory-led
approaches. The emphasis of this review will be on these four approaches of offender
profiling as their explicit aim and focus is on predicting personal characteristics, not always
the case with other forms of behavioural investigative advice (e.g., linkage analysis;
9
geographical profiling; equivocal death analysis), and will centre around the sexual
aggressors of adult women (e.g., rape and sexual murder).
1.3.1 Pragmatic Criminal Investigative Approach
In the USA, the FBI provides behavioural-based investigative and operational support
through the National Centre for the Analysis of Violent Crime’s (NCAVC) Behavioural
Analysis Unit (BAU). The BAU assists law enforcement agencies by their review and
assessment of a criminal act, by interpreting the offender’s behaviour during the crime, and
the interactions between the offender and the victim during the commission of the crime and
as expressed in the crime scene (FBI, 2008). One of the more well known examples of a
classification system for offenders is that of the Organised/Disorganised typology of serial
killers outlined by Ressler, Burgess and Douglas (1988). This classification system was
originally developed to examine lust and sexual sadistic murderers, in which ultimate sexual
satisfaction is achieved through the brutal and sadistic killing of the victim (Arrigo & Purcell,
2001). Developed through the review of case records, direct observations, and investigative
interviews with of a sample of thirty-six men who were representative of this group of
sexually oriented murderers, this dichotomy has since been applied to non-sadistic, non-serial
sexual homicides and also types of arson (Ressler et al., 1988).
According to the typology at the basic level, the organised murderer is believed to
lead an orderly life, which is reflected in the crimes he commits. He has an average to above
average intelligence, which is displayed in the evidence of planning of his offences; he will
bring and take with them a weapon of choice, and will exert control over his victim by use of
restraints, and he will be socially competent. The disorganised murderer will have average to
below-average intelligence, be socially incompetent, will display little if any planning of the
offence, and will leave the body in open view; their crime scene will show an overall sense of
disorder. Two further categories of “mixed” and “sadistic” were later developed by Douglas
et al. (1992). The mixed category contains elements from both the organised and disorganised
categories, while the sadistic category described those offenders who derive pleasure and
gratification from causing suffering and pain, through torture and humiliating their victims.
While only recently empirically tested, and found unreliable (see Canter, Alison, Alison, &
Wentink, 2004), the organised/disorganised dichotomy has been widely used and praised in
police investigations across the USA and other countries worldwide (Snook, Cullen, Bennell,
Taylor, & Gendreau, 2008).
10
Another typology of serial murder put forward by Holmes and Holmes (1998) was a
development of an earlier typology of Holmes and De Burger (1988). Holmes and Holmes’
typology outlined five classifications of serial murderers, developed through the examination
of 110 known serial murderers, through court transcripts, interview data, case studies, clinical
reports and biographical accounts. The five types in their classification were: 1) the Visionary
killer, 2) the Mission killer, 3) the Hedonistic-Thrill killer, 4) the Hedonistic-Lust killer, and
5) the Power/Control Oriented killer. The visionary killer murders because they are told to by
the visions or voices they see and hear. Their offences tend to be chaotic and disordered.
While the mission murderer kills those individuals they have judged as unworthy or
undesirable. Their offences are swift, with no premortem or post-mortem activities. The
hedonistic-thrill killer murders for the pleasure and excitement of the kill, which is often a
long process. Whereas, the hedonistic-lust killer kills for the sexual gratification, both while
the victim is alive and after they have been killed. Both subtypes of the hedonistic killer plan
and organise their offences. These killings focus on sexual gratification and sadistic acts. The
fifth type of killer, the power or control killer is motivated by the need for power and
dominance over another person, and they gain greater gratification the longer the offence
goes on (Canter & Wentink, 2004). Holmes and Holmes’ types are not mutually exclusive,
although they claim that each offender’s behaviour will have a dominate theme that would
relate to their background characteristics and from this they would be able to be classified
into a distinct category (Canter & Wentink, 2004). While, the Holmes and Holmes
classification system may use different variables and words to describe the crimes and
offenders, it is largely influenced by the original FBI organised/disorganised typology
(Canter, Alison, Alison, & Wentink, 2004).
Knight and Prentky (1990) classification model of sexual offenders is primarily based
on the motivation of the offender and takes into account that many offenders may not fit into
a discreet number of limited categories. Their classification model, The Massachusetts
Treatment Center Rape Classification System (MTC:R3), is based on the assumption that
while sex offenders are a heterogeneous group, there will be some similarities in those
offenders who commit sexual assaults (Knight, 1999; Robertiello & Terry, 2007). Based on
the examination of clinical and criminal files, standardised tests, clinical interviews, and self-
report measures, The MTC: R3 includes four typologies: the opportunistic offender (low/high
social competence) whose offences are impulsive and unplanned predatory acts, with
immediate sexual gratification as the motivating factor; the pervasively angry offender, who
11
are motivated by anger and hatred, and will use violence regardless of victim resistance; and
the vindictive offender (low/high social competence) motivated by power, control, and
hatred, who are likely to physically harm, humiliate, and degrade their victims. The sexual
offenders are subdivided into non-sadistic (low/high social competence) and the sadistic
(fantasy/nonfantasy), both are preoccupied with sex and aggression, as well as physical
inadequacy (Goodwill, Alison, & Beech, 2009; Robertiello & Terry, 2007). The MTC: R3
has been found to be a valid and reliable classification system for studying and classifying
sexual offenders (Fargo, 2007; Knight, 1999), and a valuable framework in devising and
providing treatment programmes for offenders (Canter, Bennell, Alison, & Reddy, 2003;
Knight, 1999).
Ressler et al. (1988) have outlined the stages of generating a criminal profile used by
the FBI. Stage one, profiling inputs, is about gathering and studying all the information that is
relevant to solving the crime (e.g., crime scene information, victimology, forensic
information, police reports, photos). Any information that deals with possible suspects should
not be examined or included; as such information may unconsciously prejudice the profile
and distort the impartiality and objectivity of the profile. Stage two is the decision process
models in which all the profiling inputs are organised and arranged into significant patterns. It
is during this stage that aspects of the type of homicide (e.g., single, double, triple, mass,
spree, serial), the primary objective of the offender (e.g., whether homicide was primary or
secondary motivation), the victim risk level (e.g., victim age, life style), and the risk of
apprehension for the offender are being evaluated. The levels of escalation, the amount of
time for the committing of the crime and location factors are also assessed during this stage.
Stage three, crime assessment, involves the profiler reconstructing the sequences of events of
the crime to establish just how certain things happened, how the people involved interacted
with each other and to determine which category the crime fits into, organised vs.
disorganised. The offender’s motivation is considered at this stage and combined with the
overall assessment of the crime scene. The fourth stage is the generation of the criminal
profile. The background information, physical characteristics, habits, beliefs and values, pre-
offending behaviour will be included and commented on based on the crime scene
information provided. It is at this stage that investigative recommendations might also be
made. The fifth stage in profile generation is the application of the profile to the
investigation. The criminal profile is written into a report, provided to the agency and added
into the investigation. The profiler will re-evaluate the profile if or when new information
12
becomes available. In the sixth and final stage, apprehension, the profile is evaluated for its
accuracy and success at identifying the suspect.
1.3.1.1 Critique of the Early Criminal Investigative Approach
The original profiling conducted by the FBI, based on 36 interviews, was shaped by intuition,
educated guesswork, and the agent’s experience in criminal investigations (Holmes & De
Burger, 1988). The typologies were not quantitatively tested nor based on stringent
methodological research and researchers have subsequently found deficiencies within the
profiles the FBI agents were producing (Alison & Canter, 1999; Muller, 2000).
The organised/disorganised dichotomy was based on a very small sample of
interviewed sexual murderers and lacked any comparison or control group (Coleman &
Norris, 2000). Canter et al. (2004) points out that the interviews relied on retrospective self-
reports from the offenders which can be very inaccurate as they relied on the offender’s
memory about specific points in time and on ‘trust’ that the offender did not lie about their
experiences and offences. In addition, no comparison group was used, thereby calling into
question whether any of the variables are actually specific to adulthood sexual murder
perpetration. As well, the majority of the sample used by Ressler, Burgess and Douglas did
not experience social deviance in the early years of their lives, which has been found
frequently in the backgrounds of sexual homicide perpetrators (Meloy, 2000), meaning that
any conclusion drawn using this dichotomy will not be generalisable across any other sample
or study. The original sample of men were identified as either organised or disorganised not
based on any scientific research or theoretical underpinning but on the combination of
experience and intuition of the officers involved in conducting the study (Muller, 2000). This
intuitive separation was done a priori, before any statistical tests were used to analysis the
differences between the two groups, which some argue led to a self-fulfilling prophecy, rather
than a valid behavioural dichotomy (Kocsis, Irwin, & Hayes, 1998).
Canter (1994) also criticised this classification as the boundaries between the two
distinctive typologies of serial murders are often blurred and non-distinct – many offenders
would be a hybrid of more than one type (Canter et al., 2004). Neither group is particularly
rich in detail, nor does the typology address key issues relating to the offender’s identity, nor
lend to the apprehension of the offender as they often leave the investigator with some
abstract notion of the offender and the crime (Keppel & Walter, 1999). No published
literature exists detailing just how these typologies are to be used in the evaluation of a crime
13
scene nor for the purposes of developing a profile (Palermo & Kocsis, 2005).The inability to
confidently and consistently assign offenders to one or the other type of offender affects the
ability to draw concrete conclusions about the offender’s characteristics, thereby, questioning
the pragmatic utility of this classification system.
Holmes and Holmes serial murder typology has also come under scrutiny. Hicks and
Sales (2006) have questioned the reliability and validity of the four main types as there is no
indication of any theoretical or empirical derivation. Canter and Wentink (2004) had five
major criticisms of Holmes and Holmes’ serial murder typology. The first criticism is the
lack of any systematic account of how the interviews with 110 serial murders were
conducted, and how these interviews led to their classification system. Secondly, there has
been no direct empirical testing of the five typologies (until Canter and Wentink’s) and
therefore no verification of co-occurrence of any type. The terminology used to describe each
typology is not fully described (i.e. act-focused versus process-focused) leading to
uncertainty as to under what conditions and offender or offence should be assigned to one
type or another. A further criticism was the overlap of features between the five typologies
(i.e. controlled crime scene, body movement, specific victim were listed for both lust and
power/control killer). The fifth criticism, is based around the inherent assumptions of a
typology which Holmes and Holmes’ typology fails to adhere to: “with each type, the
characteristics that define that specific type are likely to co-occur with one another with
regularity...and specific characteristics of one type are assumed not to co-occur with any
frequency with the specified characteristics of another type” (Canter & Wentink, p. 493).
Upon testing the five types using a multidimensional approach they found little evidence to
support the distinction between the serial murders based on Holmes and Holmes 1989
typology.
Knight and Prentky’s classification is not without its potential limitations. The
interpretation and classification of the offender into one of the types in the MTC: R3 is partly
subjective and based on the interpreter’s experience, skill, and intuition, potentially leading to
more unreliability (Goodwill et al., 2009). There are also concerns surrounding the
generalisability to a wider population as the typologies were developed using only those
offenders held within the MTC, which are a sample of “sexually dangerous” offenders and
therefore not representative of other samples of sexual offenders (Barbaree, Seto, Serin,
Amos, & Preston, 1994).
14
The publication of an FBI profiling methodology, described in both in Douglas et al.
(1986) and Ressler et al.’s (1988) has been criticised for lack of description about how they
constructed their typologies (e.g., organised/disorganised) (Beauregard & Proulx, 2002;
Canter et al., 2004), the typical occurrence of many of the organised features in most serial
murders, (Canter et al., 2004), and for its lack of theoretical backing (Canter, 1994; Muller,
2000). Muller (2000) also points out, the FBI methodology falls short of being scientific
based on its lack of falsifiability (the ability of a theory to be tested – verified or falsified)
and ability to propose hypotheses that are empirically testable. As much of the early FBI
methodology was based on their experience and intuition, was not part of the public domain,
and focused on perceived fantasies and sexual motivations, it was hard to empirically test.
Despite its critics, the FBI model of profiling (Canter et al., 2004; Hicks & Sales, 2006) still
remains influential.
1.3.2 Theory-Led Approach
While the original profiling and reports produced by the FBI might have been more
experience led, other endeavours incorporated theory, by trying to address the behaviours,
motivational continuum and the effects of learning on the offender (e.g., Fisher & Beech,
2007). These tried to address the criticism of the pragmatic approach of not being scientific
(falsifiable) by producing and submitting their works into the criminal investigative approach
to be peer-reviewed. Sex and aggression have been two categories of motivating factors that
have been used to categorise rape (Cohen, Garfalo, Boucher, & Seghorn, 1971; Cohen,
Seghorn, & Calamas, 1969). Groth, Burgess, and Holmstrom (1977), and Groth and
Birnbaum (1979) also argue that power and anger are primary non-sexual motivations for
rape. Each of these will occur in a variety of different forms throughout the rape literature.
For example, anger and aggression may be evident in different forms of hostility, or
destructive acts, such as verbal violence, gratuitous violence, tearing the victim’s clothing,
and acts meant to humiliate the victim (Canter et al., 2003; Canter & Heritage, 1990).
Offenders driven primarily by sex may be preoccupied with sexual fantasies and sexual
gratification or pleasure (Mann & Hollin, 2007). Power as a motivation may be expressed
through behaviours that demonstrate the offender’s control over the victim and control of the
offence. These may include the use of various levels of coercion, the binding or gagging of
the victim, and actions that suggest pre-planning and preparation (Canter et al., 2003).
Sadistic aggression may be the extreme forms of these motivations.
15
The heterogeneity of the motivations behind sex offending has lead to the
development of various typologies (Groth et al., 1977; Knight & Prentky, 1990) in which
offending behaviours are interconnected with explanations of the intentions, motivations, and
inferred offender characteristics (Canter, 1996). Groth et al.’s (1977; 1979) anger, power, and
sadistic typology of rapists is one such categorisation. The Anger rapists’ offences are
characterised by physical brutality, with excessive amounts of violence and force (Palermo,
2003), while the sexual component is used as a means to express and discharge the offender’s
built up feelings of anger and rage. Often the sexual acts are viewed with disgust by the
offender and are used to punish the victim (Pardue & Arrigo, 2008). The offender will often
use a violent blitz attack, striking, beating and tearing at their victim. Alternatively, the
offender will try to gain the trust of the victim using a confidence-style approach, talking to
them and then suddenly attacking the victim. The rape is not usually fantasised about before
hand, and the attacks tend to be of short duration and impulsive or spontaneous, triggered by
some upsetting event involving a significant (female) figure in the offender’s life. The anger
rapist uses sex as their weapon and is fuelled by the motive of revenge (Groth & Birnbaum,
1979; Groth et al., 1977).
For the power rapist, the desire of the offender is to possess their victim sexually, not
physically harm them. Often, the sexual acts become a way for the offender to compensate
for their feelings of inadequacy, and becomes a way for the offender to express their level of
mastery, strength, control, authority, identity and capability (Groth & Birnbaum, 1979). The
amount of force used is only that to accomplish the sexual assault and gain control over the
victim. Often this is done using verbal threats, intimidation with a weapon or physical force
when needed. The victims of power rapists are often held for longer periods and repeatedly
assaulted while held captive, further enforcing the idea of power and dominance over the
victim.
The sexual attacks of the power rapists are often fantasised about before hand, with
the victim initially resisting and then in spite of themselves, becoming less resistant and more
receptive, and even gratefully submitting, to the offender’s sexual prowess and embrace
(Groth et al., 1977). The fantasised excitement, anxiety and anticipated pleasure is never fully
realised for the offender and the offender is disappointed and finds little sexual satisfaction
from the actual assault. The disappointment in the offence often leads to an escalation of
violence used during the attacks as the offender becomes more desperate to achieve the
fantasised experience that escapes them (Groth & Birnbaum, 1979). The victims of power
16
rapist tend to be approximately the same age as the offender or younger and where choice is
based on availability, accessibility, and vulnerability (Palermo & Kocsis, 2005).
In the third type, the sadistic rapist, sexuality becomes fused with aggression in a
manner that transforms anger and power into something that becomes erotic, although
interconnected and often at the extremes of the various motivations and corresponding
behaviours (Canter et al., 2003). The offender finds the maltreatment of their victim
gratifying, and derives pleasure in their torment, anguish, distress, helplessness and suffering
(Groth et al., 1977; Hazelwood & Burgess, 1987). Bondage, torture and various bizarre and
ritualistic actions are the focus of the attacks, accompanied by explicitly abusive acts (e.g.,
biting, burning with cigarettes) and in extreme cases mutilation of specific areas of the
victim’s body (e.g., breasts, genitalia, buttocks) all of which play a part in his masturbatory
fantasies. In addition, foreign objects may be used to penetrate the victim sexually. The
excitement for the sadistic rapist comes from the infliction of pain upon their victim, which is
meticulously planned beforehand. This type of rapist stalks, abducts, abuses, and even
murders their victims, taking precautions against detection (Groth & Birnbaum, 1979). The
infliction of pain can provide gratification, or can act as a necessary step to other forms of
sexual activity. The sadistic rapist’s excitement increases with the level of aggression they
use, and the more powerful they feel. Typically, there is an increase in the aggression from
one offence to the next, even though the attacks themselves may have many similar aspects.
These individuals are usually able to hide their offending, and are often described as quite
personable, likeable and friendly (Dietz, Hazelwood, & Warren, 1990; Groth & Birnbaum,
1979).
The original 1977 rapist typology of Groth, Burgess, and Holmstrom was reviewed
and modified by Keppel and Walter (1999) which now includes four types of sexual murder
(see Table 1.1):
17
Table 1.1
Keppel and Walter’s Classification of Sexual Murder by Motivation, Victim Selection, and
Level of Planning
Motivation Victim Weapon Level of Offender
Selection
Selection Planning Characteristics
and Use
Power Sexual Specific; Escalation Planned Mid 20’s; criminal
Reassurance gratification may be of weapon rape; history (sexual
acquaintance selection unplanned offences); socially
and use murder isolated; unmarried;
troubled Military
service; use of porn
Power Power, Stranger Knife; Planned Early 20’s; heavy
Assertive control, rope; rape, use of alcohol and
dominance Brought to unplanned drugs; criminal
scene and murder history (burglary,
used theft); unsuccessful
relationship history;
social isolated;
troubled Military
service; antisocial;
use of porn
Anger Anger Specific; Fists; Planned Mid to late 20’s;
Retaliatory symbolic of blunt rape; unsuccessful
person they objects; planned relationship history;
seek to take knives murder criminal history
revenge on (violent offences)
Anger Sadistic Specific; Ropes; Planned Variable age;
Excitation sexual symbolic of ligatures; rape; potentially married;
(sadistic) gratification offender’s knives; planned unmarked Military
fantasies specialised murder service; use of porn;
tools of potential drug use
torture
1) Power Reassurance, represents a planned single rape of the victim with an

unplanned death resulting because the victim resists the offender’s attempts of sexual
seduction. The sexual offence is a means for the rapist to express his sexual competence and
when this fails, the subsequent killing permits him to sexually explore further and re-affirm
his sexual competence by allowing him to carry out sexual acts that he was unable to while
the victim was alive. Often the victim is someone that the offender watches, or they may be a
casual acquaintance (e.g., neighbour) (Keppel & Walter, 1999).
2) Power Assertive, this is a planned rape with little concern shown for the victim.
The killing of the victim is unplanned, often resulting from an increase in physical aggression
18
used to control the victim. The sexual assault serves as a basis to assert the perpetrator’s
masculinity and dominance over the victim (Keppel & Walter, 1999; Pardue & Arrigo, 2008).
Often there are multiple antemortem rapes of the victim. The resulting death of the victim
represents for the offender the success of asserting their power and control. The victim will
often be a stranger and was chosen based on opportunity and surprise. The offender will
spend very little time with the victim once death has been achieved (Keppel & Walter, 1999).
3) Anger Retaliatory, the rape and murder are both planned with the murder involving
overkill. This attack is perpetrated out of vengeance, and the victim is symbolic of the person
the offender is seeking to take revenge upon (Keppel & Walter, 1999).
4) Anger Excitation (Sadistic), these sexual attacks and subsequent murder are
planned. The infliction of pain, mutilation, and the terrorising of the victim is for the
gratification of the offender; all serving to feed the offender’s appetite for killing. There is
prolonged contact with the victim, who is chosen based around the offender’s fantasies,
which can last hours or even days (Keppel & Walter, 1999).
1.3.2.1 Critique of the Theory-Led Approach
While the above approaches highlight possible motivations for sexual offending, such as
power, anger and sadistic pleasure, attributed from the crime scene analysis, the classification
lacks empirical support and evidence (Fisher & Beech, 2007). To be able to infer statistical
associations there needs to be in place a system of analysis and measurement, this is where
the classifications within this approach fall down – they provide descriptions of abstract
concepts (e.g., anger; power), but do not provide a concrete way of measuring these concepts
(Cheshire, 2004). The original typology developed by Groth et al. (1977) was not developed
to specifically categorise an offender into one type or another, but as an explanation of the
different facets of a sexual offender. Nor were the types mutually exclusive or consistent,
again making it extremely hard to operationalise the different types. There is considerable
overlap between the different types of Keppel and Walter’s typology with regards to type of
crime scene (e.g., organised versus disorganised), whether the offender has previous criminal
histories, or had served in the military, their use of a weapon during their offence, and
whether they have or had emotional or relational problems (Hicks & Sales, 2006). Therefore,
the ability to draw conclusive inferences about the offender from the information from the
crime scene is not feasible.
19
While, the descriptions provided within this approach attempt to incorporate theory
around motivations and possible links with personality disorders, they are often still based on
individual experience and knowledge of those developing the classifications, and as such are
subject to the same limitations as the ‘Early Investigative approach’, as well as those of the
Clinical approach outlined below.
1.3.3 Clinical Approach
This approach to profiling is heavily reliant on clinical judgment, training, knowledge,

experience, and/or intuition, with the methods used varying according to the individual
practitioner (Alison et al., 2010). The primary focus is on the specific details of each
particular case. Profilers of this approach see each case as unique and believe they should be
treated as such (Boon, 1997). As a result, this individualistic clinical approach leaves very
few models to assess its scientific merit (Muller, 2000).
The psychodynamic approach to profiling is based on the clinical experience of the
practitioner and is necessary to make accurate prediction (Turco, 1990). Turco’s four step
model was based around the notion that all violent behaviour was a manifestation of the
mother-child struggle, where female victims were representations of all the negative elements
of the mother. In the first step, the profiler considers the crime in its entirety, looking for the
underlying psychodynamic processes. In the second step, the crime scene is assessed for any
signs of a neurological or brain disorder. Thirdly, the profiler is required to analyse the crime
scene in terms of the separation-individuation phase of the offender. Lastly, in an attempt to
construct and compose a profile of the unknown offender, the demographic characteristics of
the offender and victim are analysed.
In an attempt to set out a more systematic approach to profiling, in collaboration
Copson, Badcock, Boon and Britton compared their individual methodology and produced a
“series of steps and set[s] of features, principles and dangers which...other clinical profilers
might care to subscribe to” (1997, p. 14). What developed out of this meeting was a 10-step
procedural model (see Figure 1.1), with the centrepiece of the model being the inference of
motive, which is seen as the key to understanding the offender. The inferred motive “allows
the importation of factors from relevant literature as starting points for the development of
suggested offender characteristics” (p. 15).
The principles of clinical profiling as described by Copson et al. (1997, p. 16) instruct
that “each piece of advice should be:
20
(1) Custom made: the advice should not rely on the recycling of some kind of generic
violent anti-social criminal stereotype;
(2) Interactive: at a range of levels of sophistication, depending on the officers’
understanding of the psychological concepts at issue; and
(3) Reflexive: the advice should be dynamic, in so far as every element has a knock-on
effect on every other element, and evolving, in that new information must lead to
reconsideration not only of the element(s) of advice directly affected but of the
construct as a whole.” (p. 16).
Inherent in the principles and approach of clinical profiling are the subsequent dangers. The
desire to please may lead to an undermining of objectivity, while the close interaction
between the profiler and officer should be avoided in order to avoid any allegations that the
profile was developed to fit an already known suspect. As well, it is imperative that all data
and information be recorded, even though this is an extremely difficult and time-consuming
process. Related to this point, is the failure to produce a summary document of the amassed
information, thereby leaving the profile vulnerable to potential misinterpretation (Copson et
al., 1997). While, Copson et al. (1997) lay out a model to follow, it does not identify a
systematic process for the derivation of inferences as this is dependent on the individual
clinician. What they provide instead is a set of principles and dangers, which have relevance
in providing behavioural investigative advice.
21
1. Receive 3. Visit crime
scene
briefing 2. Request case
material depending on
the nature of the case
• WHAT (in minute detail) 4. Infer reconstruction

• HOW (in minute detail) of events
• TO WHOM (in minute) detail)
5. Infer Allows importation of factors from

motive relevant literature
6. Develop psychological 7. Introduce

constructs relating back to demographic and
what/how/to whom social factors
8. Generate a range 9. Discuss with 10. Produce

of elements of investigating officer report
advice with
probability markers
as appropriate
Figure 1.1. Ten Step Procedural Model (adapted from Copson et al., 1997)
A descriptive example of the clinical approach, as applied to sexual murderers, is the work of
Clarke and Carter (2000), who identified four types of sexual murderers through their work
with a sample of UK sexual offenders in a specialised treatment centre in Brixton Prison in
London, UK. Their profiles for types of sexual murderer were as follows:
22
1) Sexually motivated murderer, engages in sophisticated and detailed masturbatory
fantasies of killing unknown but specifically targeted victims, and who can be clearly
seen as the sadistic type with a primary motivation to kill;
2) Sexually triggered murderer, who commits an aggressive, yet controlled murder,
which uses killing as a means to keep the victim quiet and to avoid later detection.
3) Grievance motivated murderer, who commits an aggressive and uncontrolled murder
but who has no prior intent to kill, yet does so because of something the victim does
or says during the assault. Extreme violence and/or humiliation against the victim,
usually taking a sexual theme (e.g., mutilation to the genitals), will be evident,
suggesting a loss of control.
4) Neuropsychological dysfunction sexual murderer, which was developed around the
unclear motivations of one offender who exhibited clear neuropsychological deficits,
and does not necessarily depict a group of sexual offenders.
1.3.3.1 Critique of the Clinical Approach
In the same manner that the pragmatic approach relies on practical experience, knowledge,
and intuition, so do those adhering to the clinical approach. This approach is primarily based
on the individual clinician’s experience and knowledge gained through working with
individual clients, and the application of this to drawing conclusions or inferences from crime
scene information. Copson et al. (1997), and to some extent Turco (1990), provide the
building blocks of providing investigative advice, yet they fail to explicitly provide guidance
on how one would actually produce a profile. The difficulty is how “to judge when and how a
clinician’s tacit knowledge gets translated into formalized, explicit, and falsifiable
knowledge, as well as how this knowledge subsequently leads to the generation of useful
offender profiles” (Alison, Goodwill, Almond, van den Heuvel, & Winter, 2010, p.118). This
is a limitation born directly out of the fact that the inferences in practitioner driven profiling
are made through the knowledge and experiences of the particular clinician (Alison,
Goodwill, & Alison, 2005). This not only effects the ability to compare this approach with
other approaches but also the ability to compare within the clinical approach itself (e.g.,
between cases).
23
Another issue with this approach, as well as the pragmatic/theory led approaches, is
related to the Barnum 1 or the Forer effect 2 (Forer, 1949). People often assume the description
provided of the sexual offender is based on a psychological assessment procedure, even if
one has not been provided, and are therefore more inclined to accept it (Snook, Cullen,
Bennell, Taylor, & Gendreau, 2008). Many of the profiles provided by these approaches are
ambiguous and appear to describe any suspect (Alison et al., 2003; Alison, Smith, et al.,
2003). Related to this personal validation effect, is the suggestion that exposure to ambiguous
descriptions may increase the faith in psychological assessment methods and the perceptions
of the individual clinician’s views, even if the method is not valid or the profiler is not
actually skilled (Snook et al., 2008). The ambiguous nature of many of the profiles can also
be seen to support a confirmation bias 3, in which those using the profile may ‘notice’ or look
for information contained within the profile that confirms their preconceptions or hypotheses.
This has obvious implications if a criminal is then later apprehended, as any ambiguous
information contained within the profile, may appear to retrospectively describe them (Snook
et al., 2008).
1.3.4 Statistical Approach
The statistical/research approach to criminal profiling was pioneered by Canter (e.g., Canter,
Bennell, Alison, & Reddy 2003; Canter & Heritage, 1990; Canter, Hughes, & Kirby, 1998;
Canter & Ioannou, 2004). The statistical approach which asserts to be grounded in scientific
methodology, is based on the multivariate analysis of the behavioural and other crime scene
information to infer the characteristics, and psychological process, of unknown offenders
(Ainsworth, 2001). The predictions are derived from the analysis of the characteristics and
crime scene information of offenders who have previously committed crimes and those who
been apprehended, and contrast these to those being investigated (Snook et al., 2008). Canter,
established the field of ‘Investigative Psychology’, which emphasises the reliance on
psychological principles and advocates the use of scientific principles (e.g., falsifiability,
transparent processes, and evidence-based theory) in investigative focused research and its
application, including offender profiling.
1
“The phenomenon whereby people willingly accept personality interpretations comprised of vague statements with a high base-rate
occurrence in the general population” (Snyder, Jae Shenkel, & Lowery, 1977, p. 104).
2
Tendency for people to judge general, universally valid statements about personality as specific to themselves (Snook et al., 2008).
3
See Wason, P.C. (1960). On the failure to eliminate hypotheses in a conceptual task. The Quarterly Journal of Experimental Psychology,
12, 129-140.
24
One of the earliest statistical based studies was carried out by Canter and Heritage
(1990) in their development of a five-facet empirical classification for profiling sexual
offenders. Twenty-seven sexual offenders’ data, comprised of 66 assaults, were analysed and
used as the basis for their model. A method based on Facet Theory and using a type of
Multidimensional Scaling procedure (MDS) known as Smallest Space Analysis (SSA)
(Lingoes, 1973; 1979) (see Shye, Elizur, & Hoffman, 1994, for in-depth review of this
technique) was used to identify five facets of sexual offending based on the offenders’
behaviours during the commission of their offences. “...Facet theory is a structural theory. In
essence, it provides an approach to defining behavioural constructs and to testing hypotheses
concerning the correspondence between behavioural definitions and empirical observations
on variables representative of a construct” (Dancer, 1990, p. 367). Facets are sets of
attributes, sharing semantic or perceptual properties, representing the underlying conceptual
or semantic components of “some larger behavioural universe” (Dancer, 1990, p 367). For
facets to be meaningful, they must “characterise various aspects of the content universe, they
must represent conceptually distinct attributes of variables, and ...[they] must be mutually
exclusive and jointly exhaustive in the context of a particular universe” (Dancer, 1990, p.
368).
Non-metric multidimensional scaling (MDS) is used as a structural method to visually
conceptualise the structure of a content universe and the empirical structure of observations
on that universe (Dancer, 1990). SSA, a non-metric statistical analysis, represents the
relationship of every variable to every other variable as distances and points in a Euclidean
space, with greater similarity between variables resulting in their closer proximity in the
corresponding geometric space. The closer two points are in the space, the more likely the
variables co-occur. SSA attempts to find the space with the minimum number of dimensions
which preserves the rank order of relations (Guttman & Greenbaum, 1998). The distribution
of points on an SSA plot becomes more specific as the distance between the points increases;
generally, behaviours get more specific as they move farther from the centre of the plot and
farther away from other points. An SSA plot with a tight clustering of points indicates a
higher correlation between those points, and that the offending behaviours are more likely to
co-occur with all the other behaviours. A plot with points towards the outer area of the space
indicates behaviours, which are less likely to co-occur with all the other behaviours in the
dataset. The advantage of SSA, and non-metric MDS, is that no assumptions about the
underlying transformation function are made (Steyvers, 2002), and it can be used with ordinal
or categorical data (Jaworska & Chupetlovska-Anastasova, 2009).
25
The early work of Canter and Heritage (1990), and the five facets found (intimacy,
sexuality, violence, impersonal and criminality) was important as a first attempt to investigate
the relationship between offender behaviours and their characteristics distinct from their
inferred motives, which was a main criticism of the FBI and Clinical approaches. Since
Canter’s early work in the 1990’s there have been several more authors (e.g., Beauregard &
Proulx, 2002; Canter, Hughes, & Kirby, 1998; Kocsis, Cooksey, & Irwin, 2002; Lundrigan &
Canter, 2001; Porter & Alison, 2004; Porter, Woodworth, Earle, Drugge, & Boer, 2003;
Santilla, Hakkanen, Canter, & Elfgren, 2003; Youngs, 2004) producing academically peer-
reviewed research into many aspects of offender profiling.
More recently, Canter (1994) found that rapists’ behaviour could be defined in terms
of the role the victim plays for the offender (e.g., person, victim/object, or vehicle) in his
analysis of 105 cases of rape. This finding was based on the underlying interpersonal
interactions between the offender and the victim, which he maintains is distinct from any
motivational factors. Building on this theme of victim role, Canter, Bennell, Alison, and
Reddy (2003) suggested that rape could be classified by theme as well as by the severity and
type of victim violation (e.g., personal, physical, and sexual). The four themes of
classification were: 1) hostility, in which the offender uses aggression and violence to
demean/humiliate victim; 2) control, where behaviours are utilised to immobilise the victim;
3) theft, when the offender uses the opportunity for some instrumental gain; and 4)
involvement, which has the offender attempting to form a pseudo-relationship with the
victim. Although, almost a third of the rapes could not be classified as belonging to one
theme, and a fifth mixed group was created.
Another study looking at the classification of sexual offenders was that of Beech,
Oliver, Fisher and Beckett (2005) in their evaluation of the Sex Offender Treatment
Programme (SOTP). They classified a sample of 170 sexual offenders (112 rapists and 58
sexual murderers) into three groups according to the main motivation for their offending
using MCMI-III personality profiles. The grievance motivated offender was impulsive and
vengeful and blamed others for their actions. They had low insight and were highly
suspicious and resentful of others. The sexually motivated offenders planned and fantasised
about their offence beforehand, chose their victims and tended to believe that men were
entitled to have sex. They tended not to be particularly impulsive, hostile or aggressive and
used violence for instrumental purposes (e.g., to avoid detection). The sadistically motivated
offender, which consisted of sexual murderers only, was fascinated and aroused by sexual
26
violence, such as death and/or torture. They planned their offences, which often involved
strangulation, mutilation and post-mortem sexual activity.
Ter Beek, Van Den Eshof and Mali (2010) looked at a sample of Dutch rapists. Their
objective was to develop a statistical model that would be able to indicate the probability of
predicting basic offender characteristics (spatial behaviour, criminal history, living situation)
from observable crime characteristics, consisting of were modus operandi, victim-offender
interaction, and violence. They looked at separate crime scene variables (method of approach,
verbal behaviour, sexual behaviour, use of violence) and single offender characteristics
(spatial behaviour-distance travelled, living situation, previous convictions), and found that
their models for ‘distance’ and ‘violence convictions’ were promising. The study in general
supports the claim that crime characteristics can be used to indicate probable offender
characteristics.
1.3.4.1 Critique of Statistical Approach
While Canter and Alison, were highly critical of the FBI and clinical approaches of profiling,
labelling them ‘intuitive’ (Alison & Canter, 1999) the statistical approach is not without its
critics as well. Copson et al. (1997) make the point that statistics alone do not predict the
future, and extrapolation from them does not support the notion that the past will be identical
to the future, nor do they inherently support the underlying assumption that similar people
will do things, such as committing crime, in similar ways. The use of statistics does not
guarantee the inferences drawn will be valid or reliable, as these are assuming the data, itself,
is consisting of relevant and significant components, and that the statistics applied are
appropriate (Copson et al., 1997).
Sturidsson et al. (2006) attempted to replicate Canter and Heritage’s (1990) study and
their development of 5 theoretical elements of sexual offence behaviours using a sample of
146 unsolved, single victim, single perpetrator sexual assault cases collected in Sweden. The
motivational dimensions initially presented by Canter and Heritage using multi-dimensional
scaling (MDS) were not replicated. The lack of replication could be due to the differences
between Sturidsson et al.’s sample and Canter and Heritage’s. Also, Sturidsson et al.’s
sample were all single offence sexual offenders, whereas, some of Canter and Heritage’s
sample were repeat sexual offenders, which is a problem in itself, as any apparent structure
could be due to the consistency of these serial offenders. Although, upon review, Goodwill,
Alison, and Humann (2009) found that Sturidsson et al.’s use of MDS was incorrect as they
27
had used ALSCAL procedure in SPSS, which produces a dissimilarity matrix, as opposed to
using the PROXSCAL procedure, which can be used to produce a similarity matrix. This
resulted in the variables of high frequency being positioned around the periphery of the plot,
while low frequency variables were clustered more centrally, meaning the objects which are
positioned closer together are more dissimilar – which is inconsistent with other MDS studies
(e.g., Alison & Stein, 2001; Canter & Heritage, 1999; Canter, Alison, Alison, & Wentink,
2004; Canter, Bennell, Alison, & Reddy, 2003; Mokros & Alison, 2002), where the opposite
solution is utilised. Their solution makes it difficult to interpret, as well as to argue that their
results represent any display of co-occurrence, or similarity between variables, as these are
scattered around the edge of the plot (see Goodwill, Alison, & Humann, 2009 for a more in-
depth explanation).
There is question over the use of MDS, itself, as a statistical research method, because
replication across several studies (Canter & Heritage, 1990; House, 1997; Kocsis, Cooksey,
& Irwin, 2002) using similar variables has not been successful. Highly correlated variables
tend to distort MDS, with these clustering heavily in the central area of the plots and less
correlated variables being pushed outwards, making meaningful interpretation of the plot
problematic. As well, the inclusion of too few variables, upon visual inspection of the graph
and a latent dimension, makes determining rapist behaviour less apparent (Sturidsson et al.,
2006). The interpretations of the behavioural themes from the plots, where the dividing lines
are drawn, are both subjective and dependent on the individual researcher/profiler (Goodwill
et al., 2009). MDS has been cited as being affected by cultural differences (Kocsis et al.,
2002), the selection and quality of data included in the analysis, as well as how the raw data
was recorded and coded (Sturidsson et al., 2006). The sample or data used in the analysis, and
generalised into models, are based on a set of known offenders, and therefore, may not be
representative of all offenders (Wilson & Alison, 2004).
1.3.5 The Current Situation in the United Kingdom: Behavioural Investigative Advice
In the UK, offender profiling is currently provided to police forces by the Association of
Chief Police Officers (ACPO), and through the NCPE Crime Operations who recruit
individuals as full time Behavioural Investigative Advisors (BIAs) (Rainbow, 2007). BIAs
provide investigative support and advice which is grounded in behavioural sciences and in
theory. They are a professional group of individuals with vast experience of serious crime
and the knowledge to integrate their behavioural advice into an investigation (Rainbow,
28
2007). They have the potential to contribute to many aspects of the investigative process, not
just in the generation of an inferred list of offender characteristics. While, BIA still does
involve what is typically considered to be offender profiling (e.g., crime scene assessment,
offender and victimology, suspect prioritisation) it also involves providing investigative
suggestions, interview advice, risk assessment, media advice, and familial DNA prioritisation
(Rainbow & Gregory, 2009). This broader definition recognises the wider range of evidence-
based methods by which psychologists might provide advice with regards to various aspects
of a criminal investigation (Alison, McLean, & Almond, 2007).
The use of BIAs in investigations does have some limitations. While, the involvement
of BIAs in cases of serious crimes has shown to be beneficial, their involvement is best suited
to crimes where sufficient offender behaviour is evident and where sufficient discrimination
exists between offenders within a certain crime type (Rainbow & Gregory, 2009). As well,
BIA reports only provide the most likely type of individual, there may be significant variance
found in a minority of cases with regards to the reported prioritisations, and do not provide
information regarding the guilt or innocence of an individual. The advice provided is for the
increased understanding of an event and for informing and prioritising investigative decision
making and actions (Rainbow & Gregory, 2009).
Each of the approaches has their inherent strengths and weaknesses which have been
outlined above. That said there are more general critiques of offender profiling that are
consistent across the different approaches.
1.4 Critique of Offender Profiling in General
This section will look at the pragmatic use and validity of profiles, and the quality of data
used in profile development.
1.4.1 Pragmatic Use and Validity
There are a number of general drawbacks within Offender Profiling research as a whole. Two
important concepts that need to be considered are 1) the validity of the profile, the accuracy
of predicting the characteristics of unknown offenders, and 2) the utility of the information
contained within the profile, whether it can be used pragmatically by investigators (Kocsis &
Palermo, 2007). There have only been a handful of studies, which have directly attempted to
test the validity and accuracy of profiles and the abilities of profilers as compared to non-
29
profiling groups (e.g., Copson, 1995; Kocsis, 2004; Kocsis, Irwin, Hayes, & Nunn, 2000;
Pinizzott & Finkel, 1990). Bennell, Jones, Taylor, and Snook (2006) critic the research by
Kocsis and colleagues, in the fact that they state their profiler groups are more accurate than
the comparison groups, often students, police officers, or psychologists, yet this is not in
terms of “absolute” accuracy of the predictions made by the profiler, but between the relative
predictions made between the different groups. So while, it may look like the profilers are
more accurate in their predictions, they are still only found to be accurate approximately 45%
of the time (Bennell et al., 2006). Bennell et al. argue that this level of accuracy is not high
enough to be investigatively useful, as half of the information provided was not useful, and
there is no minimum useful accuracy level set. Some of these studies suffer from internal and
external validity problems. The sample of professional profilers used is quite often low
(Dowden et al., 2007; Kocsis) and may not be representative. As well, there is a lack of an
objective and tested criteria with which to test a sample of actual profiles against (Homant
& Kennedy, 1998) as many of the studies testing the accuracy and validity of profiles are
artificial in nature (Dowden et al., 2007; Kocsis, 2003), limiting their external validity.
The FBI claim to have tested the validity of profiles composed by their Behavioural
Science Unit in an internal report, with the finding of an 80% degree of accuracy. Yet, this
report has never been made public and is only known as it is mentioned in Pinizzotto’s (1984)
work. As this document has not been made public, or made available for scrutiny, the claims
of accuracy cannot be verified. There are many examples of profiles in true-crime stories or
biographies in which claims of support of profiling accuracy are made, although the ability to
confirm the validity of the profiling techniques used or the profilers themselves is limited, as
a large number of these true-crime biographies are written by profilers themselves (e.g. ‘The
Jigsaw Man’ and ‘Picking up the Pieces’ by Paul Britton).
A study by Alison, Smith and Morgan (2010) raises cause for concern when it comes
to the perceived accuracy and usefulness of offender profiles. They found in two separate
studies that the majority of officers and forensic professionals rated both fabricated and
genuine profiles as at least somewhat (75%) accurate despite being given distinctly different
descriptions of the criminal, and the majority rated the profile as useful. This relates back to
the perceived accuracy and utility of offender profiles mentioned earlier in this chapter.
Regardless, of the mixed support for the accuracy of offender profiles, their usefulness as an
investigative tool is also in debate. Some studies show that investigating officers utilise
profiles because they believe they work, and are useful in identifying and prioritising suspects
(Copson, 1995; Jackson et al., 2003). For those officers that do not necessarily believe that
30
profiles are useful, yet still use them, this could be because they feel there is nothing lost by
using all available investigative techniques (Snook et al., 2008).
1.4.2 Quality of Data
The quality of data used in profiling research is often limited by what is available, how it can
be coded and how rich and/or robust it is. The type of data most often utilised in offender
profiling research comes from law enforcement agencies where there are variations in
collection protocols across the different agencies (e.g., no systematic guidelines for
information collection; time constraints), resulting in lowered internal validity. The evidence
is collected for the purpose of a police investigation, not for research purposes, and often with
little contextual grounding or concern with its quality or reliability (Alison, Snook, & Stein,
2001). The result is low levels of validity, especially external validity, which affects the
generalisability of any results (Dowden et al., 2007). Without the ability to generalise, and
extend the results obtained from studies and make predictions about the larger population of
offenders, the utility of offender profiling is called into question. For example, the original
FBI organised/disorganised dichotomy may have been accurate in so much as it explained
some of the facets of the 36 individual sexual offenders on which it was based, but it severely
lacked any external validity, or the ability to be used to predict behaviours or characteristics
of offenders not included in the original sample. A similar argument can be made with Groth
et al.’s (1977), or Clarke and Carter’s (2000) or Canter and Heritage’s (1990) classifications.
Many studies have limited sample sizes due to the nature of the data collection. While
the majority of those with access to the appropriate data are Law Enforcement agencies, the
majority of the research is often done by professionals in the academic world. There are many
data protection issues, which must be addressed for academics to gain access to appropriate
data. Even after access has been granted, the quality of the data still must be taken into
consideration. These data can only be as good as what is available to the officers at the time
of the investigation. What is collected is often inaccurate and not in a form that is conducive
for empirical research (Mokros & Alison, 2002). Equally, the main advantage in using
evidence collected during an investigation is that it represents naturally occurring behaviours
exhibited by an actual offender; not a controlled subject in a controlled laboratory (Alison et
al., 2001). At present, the effective utilisation of the data requires careful consideration of the
biases potentially inherent, and any conclusions drawn from research must keep these
limitations in mind when making generalisations (Alison, et al., 2003).
31
1.5 Summary
Is offender profiling effective? An important question when one considers the faith in profiles,
and the status it is given in forensic investigations (Snook et al., 2008). While, some of the
preceding studies may have limited results, they show there is the possibility of inferring
some offender characteristics from crime scene behaviour and providing (albeit limited)
support to the underlying assumptions of offender profiling. The area of Offender Profiling
generates a lot of interest in both the academic field and the everyday world as a result of a
few highly prolific cases (e.g., Jack the Ripper, Boston strangler). Historically profiling has
been based on ‘intuition’ and experience, but as the field of Offender Profiling has matured
the need to be more scientific in approach has led to the development of models/typologies of
offender behaviour based around the findings of empirical studies. Different approaches have
attempted to define, and operationalise offender profiling based on the individual principles
inherent in the approach. The Criminal Investigative approach initially relied heavily on
intuition and experience of the FBI agents and who researchers who started publishing in the
area of profiling. Although, the development of large databases and systems containing
information on serial and violent crime/criminals, such as the FBI’s Violent Criminal
Apprehension Program (ViCAP) (Collins, Johnson, Choy, Davidson, & MacKay, 1998;
Howlett, Hanfland, & Ressler, 1986) and the Violent Crime Linkage System (ViCLAS) in
Canada (RCMP; Collins et al., 1998), has allowed for the utilisation of a lot of information
and data, and the drawing upon of many of the same theories and models that many in the
academic field and other areas of investigative psychology use (Snook, Luther, House,
Bennell, & Taylor, 2012). The Clinical approach developed a model of offender profiling
centered on the concept of motives. While, the Statistical approach aimed to provide a
testable psychological and scientific framework for inferring characteristics. None of the
approaches alone explain the complexities of offending. The FBI/Pragmatic approaches bring
with them a multitude of investigative experience; the Clinical an abundance of medical and
intimate client-based knowledge; while the Statistical approach provides a means in which to
more objectively measure and examine offending behaviour. Without the experience,
knowledge, and information that is engrained and gathered in the first two approaches, the
ability to know which variables to look for or code for would be lost. While the latter
statistical approach, allows for the removal of the individual, the individual opinions and
biases, and for an objective examination of the patterns and findings. Therefore, the way
forward should seek to integrate all of the approaches (Alison, West, & Goodwill, 2004;
32
Alison et al., 2010). Together the approaches strengthen each other and give weight and
support to each other and more importantly, offender profiling as a whole. This will be
further examined in the next chapter.
33
Chapter 2: Directions for Offender Profiling and Current Thesis
2. Introduction
In its simplest form Offender Profiling is the determination of offender characteristics

from the behaviour offenders exhibit and leave indication of at a crime scene. The ability to
make such inferences requires both consistency in offending behaviour and distinctiveness.
Simply expressed offender profiling is represented by Canter’s (2011) “profiling equation”,
or abbreviated as the “A  C equation”; where the inferences, represented by the arrow, are
derived from the actions, such as the crime location, time, victimology, etc., which are
informative about the characteristics of the offender (Canter, 2004; 2011). Although, this
may be better described as a decision pathway or process, as strictly speaking it is not a true
equation as the two sides do not completely equate. Though, the process can be described as
bi-directional based on the stated premise that the actions reflect the characteristics of the
offender, and the offender characteristics will also influence the actions of the offender.
Offender Profiling has been largely based upon personality and trait approaches, which see
the basic units of personality, and therefore, behaviour, as largely non-situational and based
on context-free dispositional constructs (Alison, Bennell, Mokros, & Ormerod, 2002; Davies
et al., 1997; Pervin, 2002). There is evidence that behaviour can be predicted across situations
based on scores of basic trait dimensions, often cited in employment and job performance
literature (Hogan, 1991; Hogan & Ones, 1997). One explanation is that once self-schemas
become well organised, through our experiences, we selectively respond to information in
ways congruent with our expectations and self-views (Fiske & Taylor, 1991). With regards to
longitudinal stability, there is moderate support for the stability of temperamental
characteristics observed in childhood and personality in young adulthood (Caspi & Silva,
1995), even connections with childhood temperament – measured as early as age 3- and
criminal behaviour (Block, Block, & Keyes, 1988; Caspi, 2000; Raine, Reyonds, Venables,
Mednick, & Farrington, 1998). However, there is greater support for stability of personality
throughout adulthood (Alwin, Cohen, & Newcomb, 1991; Glenn, 1980; Pervin, 2002). As we
age we become more consistent and less likely to change, but there still remains the potential
to change (Roberts & Del Vecchio, 2000).People unconsciously filter experiences, elicit
responses from others, and choose certain life paths which are consistent with their
personality, so even the person-environment interactions can be seen to support a level of
34
continuity in personality, as well as allowing for the potential for change (Caspi & Roberts,
2001).
There is a distinction between personality and a personality disorder (PD) that should
be noted, although both will influence behaviour and the understanding of that behaviour.
Personality is a set of distinctive traits and characteristics that distinguishes an individual, or
a nation or a group and encompasses fairly consistent patterns of thoughts, feelings, and
behaviours (Merriam-Webster). Since Cattell’s early work (1943; 1945a b), the ‘Big Five’
factors (openness, conscientiousness, extraversion, agreeableness, and neuroticism; OCEAN)
have been found and replicated by various researchers (e.g., DeRaad, Perugini, Hrebickova,
Szaroca, 1998; Digman & Takemoto-Chock, 1981; Norman, 1963; Saucier & Goldberg,
1996). After a dormant period they were developed into the taxonomy that it is today by
Costa and McCrae (1992) 4.
According to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5)
(American Psychological Association, 2000) personality disorders are enduring, longstanding
maladaptive patterns of perceiving and responding to other people, and behaviours associated
with significant distress or disturbance in self and interpersonal functioning. Where
personality and PDs merge can be seen in the different criteria for the PDs as these are linked
to personality and personality traits (Costa &McCrae, 1990). Borderline and avoidant PDs
have been found to be associated neuroticism from the five-factor model of personality, and
antisocial, paranoid, and narcissistic disorders have been linked with low agreeableness.
While, schizotypal PD has been found to be neurotic and introverted (Costa & McCrae,
1990). Linking personality, PDs, and offending, it’s been found that avoidant, schizoid, and
dependent tendencies have been associated with sadistic rapists, narcissism, paranoia, and
antisocial tendencies with opportunistic offenders (Proulx, St-Yves, Guay, & Ouimet, 1999),
and violent sexual offenders, as opposed to less violent sexual offenders, tend to score higher
on histrionic, narcissistic, antisocial, and paranoid tendencies (Langevin, Paitich, & Russon,
1985; Proulx, Aubut, Perron, & McKibben, 1994).
The relationship from crime scene aspects to offender characteristics is not as straight
forward as it initially appeared to be (Goodwill & Alison, 2007; Mokros & Alison, 2002).
The lack of substantial support for the assumption of homology, and the imperfect support
found for behavioural consistency is evidence of this – there are other influencing factors not
4
See John, O.P., & Srivastava, S. (1999). The big five trait taxonomy: History, measurement, and theoretical perspectives. In. L.A. Pervin,
& O.P. John (Eds.), Handbook of personality: Theory, and research (2nd edition) (pp.102-138). New York: Guildford Press, for a more in-
depth summary of the development of the Big Five.
35
being considered or included in this relationship. A better definition of offender profiling
might be “the application of psychological theory and behavioural evidence analysis to the
investigation and reconstruction of physical evidence that relates to a particular offender’s
crime scene characteristics, victimology, motivation and behaviour patterns” (Gee &
Belofastov, 2007; pp. 62). However, even this definition leaves out key aspects of
information that many profiles overlook – offender perception and offence context. The
influence of such factors is recognised in models of personality (e.g., Mischel and Shoda’s
[1995] CAPS model).
Traditional personality functioning is seen as more of a static consistent model, while
more current conceptions of personality and behaviour, see behaviour as more dynamic and
conditional on the individual and the specific situations in which they find themselves
(Shoda, Mischel, & Wright, 1994). The debate is whether people show consistency in their
behaviour across situations because of inherent internal personality traits (the person side of
the debate), or whether behavioural consistency is affected by the situation, is not inherent in
the person, and therefore is contextually variable (the situation side of the debate) (Pervin,
2002). While this debate has not completely dissipated, there has been some headway in an
attempt to reconcile the two camps. Allport, Cattell, and Eysenck (founding influences of
personality psychology) all recognise to some degree the importance of the situation and the
variability of individual behaviour. Allport (1961) acknowledged that our personality
dispositions are never completely consistent, but also empathised that we do exhibit relatively
enduring patterns of thoughts, feelings, and behaviour (Roberts & Caspi, 2001). Personality
interactionists believe the more important question is “how do characteristics of the person
interact with characteristics of the situation”(Pervin, 2002, p. 78) which allows for the
multiple factors that contribute to the exhibition of behaviour (Ahadi & Diener, 1989), and
for the interaction between both the internal personal traits, and the influence of the situation
on a person’s behaviour.
The link between A and C represents a complex and challenging set of variables and
circumstances that work to modulate criminal behaviour – a relationship largely ignored in
the current pool of literature. The original idea that the way a person thinks directs the
person’s behaviour (Douglas et al., 1986) does not account for the multitude of other
variables that are present at any given moment which influence behaviour. No two offenders
are exactly alike; they do not do the very same thing in the very same way or for the very
same reasons throughout their offences. Although, if the situation in which the offender’s
crimes are committed are similar, and have the same or similar psychological meaning for
36
them, it would follow that behavioural consistency in this circumstance would be expected
(Mischel & Shoda, 1995; Shoda, Mischel, & Wright, 1994) 5. Even when the same discernible
behaviour is observed there can be several different reasons for why and how this behaviour
was brought about – similar acts can occur for different reasons, and different acts can
happen to serve similar purposes (Douglas et al., 1986). If this is true, there are implications
for offender profiling. While no professional will argue against the fact that people do not
respond exactly the same way in exactly the same situations, this does not mean there are not
consistencies within a person’s response. Looking at overarching behavioural domains,
personality psychology has found there is variation in the consistency of non-criminal
behaviour (Funder & Colvin, 1991; Furr & Funder, 2004), concurrently studies looking at
criminal behaviours, such as Grubin et al. (2001) and Bennell and Canter (2002), have also
found this. Thus, while there will be variations across people’s behaviours, and more
specifically in the ways offenders commit their crimes, there will also be many significant
similarities. When considering the differences and similarities among people, and offenders,
given similar characteristics and backgrounds, similar thought processes, and similar
situations, we could expect a similar degree and level of responses and behaviours to occur,
both across offenders, and within a single offender 6.
In an attempt to integrate the approaches to personality theory, and understand the
stable intra-individual patterns of variability inherent in an individual’s behaviour across
situations more dynamic conceptualisations of trait theory have been developed. One such
model is Mischel and Shoda (1995; 1998) cognitive-affective processing system (CAPS) 7.
CAPS places the conception of personality within the social world in order to contextualise
the individual and allow the examination of the reciprocal interaction between person-
environment (Mischel & Shoda, 2008). Person variables, such as how people construe/encode
situations and themselves, are important, but these are components of a dynamic and
interconnected organised system of relationships, that interacts with the social-psychological
situations (Mischel & Shoda, 2008). CAPS theory assumes that people differ in the ease with
which cognitive and affective mental representations or units, CAUs, become active, but also
that individual differences reflect the accessibility of CAUs as well as the distinctive
organisation of relationships among them (this is the stable structure of the personality
system) (Mischel & Shoda, 2008). The CAUs are comprised of constructs, expectations, and
5
See Chapter 1 for a review of the support for the Behavioural Consistency assumption.
6
See Chapter 1 for a review of the support for the Homology Assumption.
7
The CAPS model has been the focus in much Offender Profiling discussions (e.g., Alison et al., 2002; Markson, Woodhams, & Bond,
2010; Sorochinski & Salfati, 2010; Woodhams, 2012) and has been chosen as a focus of the current thesis so the findings can be related to
what has been done before.
37
beliefs of how the person sees themselves, people and events around them, and the situations
they encounter. They encompass the affects, the feelings, emotions and affective responses,
as well as, the desired outcomes and goals of the individual. The potential behaviours and the
if...then scripts and strategies for attaining outcomes, and one’s own behaviour and internal
states are also a part of the CAUs (Mendoza-Denton & Mischel, 2007). The CAPS theory
postulates that the meanings of situations will vary and have different impacts between
people, as well as within an individual, and on different occasions (Eaton, South, & Krueger,
2009). This means that different situations will activate different CAUs; producing an
if.....then behavioural contingency. Therefore, when the situations (the ifs) change, so will the
thens (Mischel & Shoda, 2008). CAPS theory highlights three points: 1) personality systems
are understood in terms of their cognitive-affective units, as well as, the coherent organisation
of those units; 2) this system functions and interacts with the social environment. Lastly, 3)
people will behave in variable and distinctive manners which characterise that individual
(Cervone, 2005).
The CAPS model is one paradigm for theoretically testing and supporting the
assumptions of Offender Profiling as it allows for both behavioural consistency within the
individual, even cross-situationally. In addition, it provides logical underpinning for the
hypotheses of behavioural distinctiveness and similarities across crime types and between
offenders.
2.1 Rationale for Current Research
Psychological theory therefore suggests the link between offence and offender characteristics
is more complex than has previously been depicted in offender profiling literature. The
cognitive-affective units, such as the perceptions and motivations of the offender, and the
context of the situation will all exert their own authority to affect the way in which the
offender ‘reaches’ their decision to offend, who the offender ‘chooses’ to offend against and
how the offence is carried out. As outlined above, the CAPS model gives a theoretical
underpinning for the reciprocal relationship between the person-environment and resulting
behaviour. While, the relationship between certain offence characteristics, offender
characteristics, and contextual variables are not always clear, the one’s chosen for inclusion
in the current study, and in the following chapters, are often those that would be present in an
offender profile of a sexual offender (see Ault & Reese, 1980; Davies, 1999; Ressler,
Burgess, & Douglas, 1988; Ressler & Douglas, 1985) and the information contained in police
38
files (Mokros & Alison, 2002). As the CAPS model stands, it provides a “general framework
for building a more cumulative, integrative science of persons interacting dynamically with
their socio-cultural psychological life situations, each reciprocally influencing the other”
(Mischel, Mendoza-Denton, & Hong, 2009, p. 1366).
Figure 2.1 represents an outline of a revised A to C equation. The revised equation
illustrates how the relationship between A to C should be seen as a non-direct relationship,
outlining the interplay between the person’s demographics, dispositions, situations and
contexts encountered. The relationship between A and C is influenced/mediated by
Perceptions, Motivations and Context, which I will now examine in more detail.
Context/Situation
Location
Drugs/alcohol
Victim Resistance
Crime Scene Actions Offender Characteristics
Psychological Make Up
Implicit Theories
Motivations
Figure 2.1 Revised A  C: Process of Offender Profiling incorporating Perceptions,

Motivations, and Context
The concepts described below as potential influencing factors provide more of a holistic
approach to offender profiling which is grounded in psychological and sociological theory. In
the absence of models examining the reasons why we should be able to infer socio-
demographic variables from crime scene characteristics and aspects so often cited in profiles,
39
these provide a building point based in theory and not just conjecture to start linking
characteristics with crime scene aspects.
2.2 Context
As outlined by the CAPS model, as well as other theories of behaviour (e.g., Routine Activity
Theory, Cohen & Felson, 1979; Rational Choice Approach, Clarke & Felson, 1993), the
context in which a crime occurs plays a significant role as to how the offence is actually
committed. Within our daily lives, we engage in legitimate activities which dictate where we
will be. For offenders, this is also mixed with the locations of their illegitimate activities, and
the locations of potential victims (Blackburn, 1993; Cohen & Felson, 1979). Criminal
violations are routine activities, which share many of the same attributes of and are
interdependent with, other legitimate routine activities. People participate in legitimate
routine activities daily to satisfy their personal needs, through work, childrearing, shopping or
leisure pursuits and it is these routine daily activities which determine where and when
people are, and what they are doing, and hence the location and vulnerability of personal and
property targets (Blackburn, 1993). When deciding to commit a crime, an offender must
decide (either implicitly or explicitly) what the various options of action are as well as their
corresponding consequences. During this process, different situational and contextual
variables can influence the consequences of alternative courses of action and thus the
decisions made and the actions taken (Cornish & Clarke, 1987; Clarke & Felson, 1993).
Therefore, context acts a mediating unit, which interacts with the personality processes of the
individual resulting in the behavioural expressions exhibited by that person (Mendoza-
Denton & Mischel, 2007). Under this approach, sexual offending cannot be understood
outside or apart from the ecology of everyday life (Sampson, 2001).
Looking specifically at the relationship between crime scene information and offender
characteristics, Goodwill and Alison (2007) found that this relationship is moderated by
other aspects of the crime, more specifically the situation and context surrounding the crime.
They found that victim age can accurately predict offender age, within three years, when the
level of planning and the use of (excessive) violence is taken into consideration as
moderating factors. Goodwill and Alison found that in a planned offence, victim choice
reveals more detail about the psychological motivation of the offender than in an offence that
has no indication of planning. The interaction between the offender and his victim reflects
another person-situation influence on offending behaviour. The offender’s response to the
40
victim’s behaviour (resistance/compliance) will be defined by the offender’s attitudes and
roles they place upon their victim (Canter, 1989; Davies, 1992). In her 1992 paper, Davies
describes a serial rapist who changed his behaviour as a result of the victims’ behaviour; with
those who had resisted more heavily he was more violent and threatening with, while those
that had been compliant, he used minimal aggression towards, and with one he even
arranged a date to see her again. Body disposal patterns of sexual murderers were also found
to be related to the interaction between the situation and the offender characteristics
(Beauregard & Field, 2008). Offenders in a relationship at the time of their offence and who
offended against strangers (excluding prostitutes) were more likely to move the body from
the offence location. Whereas, the victim’s body was more likely to be left at the crime scene
if the victim and offender had had an altercation prior to the offence, if the offence had
happened at night and if the victim was older (Beauregard & Field, 2008).
The context or situation that a crime occurs in is a highly important variable to
consider. The person may have the motivation to offend, and the acquired perceptions to
offend, but without the “appropriate” situational or contextual variables needed to necessitate
the offence behaviour, the question remains if that offence would still occur. It is important
to include the context and situation of offending because, according to CAPS, specific
psychological meanings of the situation will result in specific categories of behaviours –
meaning if similar situations are encountered one can infer that similar behaviour will also
ensue. Pragmatically, this would allow investigators the potential to predict how an offender
may behave in future offences, as well, as link previous offences to that offender based on
behavioural consistency.
2.3 Perceptions
From an early age our knowledge is organised into theories, which help facilitate our
understanding of the world and allows us to explain and understand our social environment
and make predictions about potential future events (Beech, Fisher, & Ward, 2005). Ward
(2000) suggests that the perceptions, or schemas, a person possesses are in fact underlying
implicit theories that the person holds about the world. These give control and development
to a person’s internal life, and link the individual with their social environment and give
meaning for the events of their lives (Beech, Fisher, & Ward, 2005). Simply put, the
perceptions of ourselves, and our environment, components of our CAUs, will ultimately
dictate how we react and behave in different situations we encounter (Mischel & Shoda,
41
1995). This is based on the ease with which these CAUs becomes activated as depicted by
their relevance to the given situation, resulting in subsequent behaviour (Mendoza-Denton &
Mischel, 2007).
These theories will directly influence any assumptions a perpetrator will make about
their victim’s internal states, what they predict the victim will do, and will partially affect
how they react to their victim’s responses (Blumenthal, Gudjonsson, & Burns, 1999). These
implicit theories represent our comprehension of close relationships, other people’s actions,
and the structure of our world, and the nature of mental states (Ward, 2000). They are also
influenced by the different motivational aspects of a given interaction between the person-
environment.
2.4 Motivations
Motives and motivation explain why we behave in the manner in which we do. They
influence our cognition and action, as well as, our thinking and our behaviour, and are an
important aspect of CAPS and the construals activated during the processing of social
information (Mischel et al., 2009). Motivations play a role in why we respond differentially
at various times to the same stimulus (Pervin, 2002). “The concept of motivation suggests
that there are internal qualities that play an important role in the activation and regulation of
behaviour” (Pervin, 2002, p.36).
The broadest categorisation of motivation within sexual offending is that of the sexual
versus non-sexual primary motive. Historically the most cited motivations behind sexual
offences were the sexual thrill or sexual intoxication that accompanies the sexual offence
(Scully & Marolla, 1985). In favour of an underlying sexual motivation is that if control or
domination over another person was the motivation behind sexual offending, then we should
see an equal distribution of male and female victims, at various ages, not just reproductive-
aged females who tend to be the majority category of victim for sexual offences
(Shackleford, 2002; Wilson, Daly, & Scheib, 1997). In addition, the fact that there is always
a sexual component, with the offender often reaching orgasm, is cited as another supporting
factor for the sexually motivated hypothesis (Myers, Husted, Safarik, & O'Toole, 2006) .
Other motivations such as power and control over victims, or the expression of anger or
revenge towards the victims have also been given support (Myers et al., 2006). Groth and
Birnbaum (1979) observed that in all cases of forcible rape, power, anger and sexuality are
always present to some degree, and that rape is a crime of violence. Myers et al. (2006) argue
42
the primary motivation behind sexual offending behaviours is that of sadistic pleasure. Any
other motivation (e.g., anger, dominance, sexuality) has secondary purposes – to either
increase or heighten the sexual arousal, or are of practical use in managing the victim so the
offence can be carried out.
While motivations themselves are internal constructs that cannot be themselves tested,
they can be inferred from observed concrete behaviour (Amir, 1971). They can be inferred
from the behaviour of someone with what would be expected in particular situation if that
person was motivated in a particular manner (Palmer, 1988). Motivations impel us to act,
there are motivations behind every offence committed and it is these motivations, which hold
precious insight into the mind, the thoughts, feelings, and the behaviours of the offender and
their offences, as these are an external presentation of internal constructs.
2.5 Conclusions
The concept that diverse variables influence behaviour is not a new one, and
similarities can be found between the revised A to C equation (Fig 2.1.) and an aetiological
model of risk for sexual offenders developed by Beech and Ward (2004) (Figure 2.2). The
vulnerability factors of Beech and Ward’s model are similar to the perception variables in
Figure 2.1 and the CAUs of the CAPS model. Offence-supportive cognitions, grounded in a
set of core schemas or implicit theories held by the offender, generate the cognitive
distortions that are measured at the surface level (Ward, Polaschek, & Beech, 2006). The
state acute dynamic factors can be equated with motivations, and the triggering/contextual
risk factors with the context variables. For example, the need for intimacy may motivate a
sexual offender to commit rape in an attempt to develop a pseudo-intimate relationship with
their victim (Canter & Heritage, 1990). In this case, the primary motivation for the rape is the
desire for social contact (Marshall, 1989), and the offender may ask the victim questions,
compliment the victim, kiss the victim, and even apologise for the attack, or make the victim
reciprocate in making sexual comments (Canter, 1994; Canter et al., 2003). What is evident
from the model is that all these factors interact in the process of behaviour formation and the
likelihood that a behaviour will even occur (Mischel & Shoda, 1995; Ward et al., 2006).
43
Developmental Vulnerability (trait) factors State factors
factors
Marker variables Psychological dispositions Acute dynamic
Sexual Abuse Static/Historical Stable dynamic risk Deviant sexual

risk factors factors arousal
Rejection Deviant thoughts
Persistence/range of Sexual self-regulation and fantasies
Attachment offending Offence-supportive Need for
problems Psychosocial cognitions intimacy
Level of Risk
problems Level of interpersonal Affective states
Criminal/antisocial functioning (positive or
history General self- negative)
regulation problems
Triggering
events/contextual
risk factors
Victim access
Non-cooperation
with supervision
Social dislocation
Substance abuse
Relationship
conflict
Antisocial peers
Figure 2.2 An Aetiological Model of Risk (adapted from Beech & Ward, 2004) 8
Motivation is not always straightforward or easy to infer from crime scene

information. Yet, as suggested in models such as Beech and Ward (2004), and by psychology
in general, it is too important an aspect of any behaviour to be ignored or left out. It is an
imperative piece of the puzzle when answering the question ‘Why’ – why an offender chose
to commit their crime, why they chose their victim, why they committed the crime in the
manner they did, etc. Motives are fundamentally influential to our behaviour as they compel
our thoughts and propel us into action. Intertwined with motivation is the offender’s
perception of his world. How the offender views his personal world, his relationships with his
family, friends, work, and strangers will all influence and direct how he chooses to engage
with and react to them. The contextual variables surrounding an offence are often
undervalued or overlooked. The use of alcohol or drugs prior to the offence, the location of
initial contact with victim and of the actual offence, and the level of victim resistance are just
some of the contextual factors that may prove to be invaluable in the process of Offender
8
See Beech, A.R & Ward, T. (2004). The integration of aetiology and risk in sexual offenders: a theoretical
framework. Aggression and Violent Behavior, 10, 331-63 for more detailed explanation of the model.
44
Profiling and in providing investigative advice, yet little research focuses on these factors.
The inclusion of context, the perceptions and motivations of the offender provide a more
comprehensive approach to the basic idea behind offender profiling, which if viable should
be able to more effectively generate offender characteristics from crime scene information,
thereby allowing for more specific suspect elicitation and prioritisation and greater pragmatic
use through their utilisation in the understanding of various circumstances in which sexual
offending occurs, and the use of this knowledge in police interviewing and the reconstruction
of the offence (Beauregard & Fields, 2008).
Therefore, the current thesis will look to explore the mediating relationships between
contextual variables, perceptions of the offender, and the motivations behind their offending.
The general exploratory hypothesis being that each of these constructs will significantly
influence the resulting offence behaviour and themselves link back to specific offender
characteristics.
45
Chapter 3: Methodology
The purpose of this chapter is to outline the general methodology and procedures used
throughout the studies included within the thesis. A description of the sample can be found
below, as well as, in Chapter 4, and short descriptions of the procedures and materials can be
found in each of the individual chapters. Below is a more in-depth and complete description
of the analyses used in the studies in Chapter 4, Chapter 5, Chapter 6, and Chapter 7.
3.1 Sample
One hundred and two sexual aggressors were drawn from an original sample of 170
convicted sexual offenders (112 rapists; 58 sexual murderers), who were approached before
taking part in the 55 different Core Sex Offender Treatment Program (Core SOTP) run by the
UK Prison Service between 1998 and 2002, across seven of Her Majesty’s Prison
establishments in England and Wales. Eighty-six rapists and 45 sexual murderers voluntarily
agreed to and completed the original interviews. Based on a ‘cognitive-behavioural’
approach, the Core SOTP is an intervention programme. The aim of SOTP is to increase the
offenders’ motivation to avoid re-offending by targeting and challenging their offence-
supportive cognitions serve to maintain their sexual offending behaviour by helping them to
develop new attitudes and the self-management skills necessary to accomplish this. The
average length of the programme is 180 hours, completed in approximately two-hour content
sessions, meeting for two to five sessions per week (Beech et al., 2005).
Only those sexual aggressors who had committed and were convicted of at least one
sexual offence against an adult (16 years of age and above) 9 female victim were included in
the thesis sample. Sixty-four of the offenders had committed rape as their index offence,
while the remaining 38 were sexual murderers.The term for rapist will be in line with the
Sexual Offence Act 2003, which states that:
A person (A) commits a [rape] if a) he intentionally penetrates the vagina, anus or

mouth of another person (B) with his penis, b) B does not consent to the penetration,
and c) A does not reasonably believe that B consents. (Chapter 42, Part 1, Section 1)
9
Based on the Sexual Offence Act 2003 age of consent to sexual activity being 16 years of age and older.
46
The term sexual murderer will be the same as those used by Beech, Oliver, Fisher, and
Beckett, (2005), and “applied to individuals who have killed someone where there is either
clear forensic evidence of a sexual element to the killing, or a sexual component is admitted
or suspected” (p. 5).
This sample represented approximately 1.9 % of the total male population under
immediate custodial sentence for sexual offences at that time and about 0.2% of the total
male prison population under immediate custodial sentence (Home Office, 2004). The
samples’ ages ranged from 14-57 years old at the time of the offence (M = 26.70, SD=8.60).
The mean sentence length for the total sample was 10.70 years (SD=4.35), ranging from 5 to
25 years.
Table 3.1
List of the Seven of Her Majesty’s Prison Establishments across England and Wales.
Establishment Rapist Sexual Murderers Total
Brixton 2 20 22
Albany 12 3 16
Maidstone 9 0 9
Frankland 5 1 6
Full Sutton 8 3 11
Wandsworth 15 0 15
Wakefield 12 11 23
Total 64 38 102
3.2 Chapter 4 Methodology
In Chapter 4, to compare and contrast rapists and sexual murderers, Pearson’s Chi-square
tests were calculated for the nominal data and Independent Samples t-Tests, Mann Whitney U
and MANOVA tests for the interval data. Agglomerative hierarchical cluster analysis was
used to investigate whether the offenders in the sample would form similar clusters to those
reported in previous studies.
47
3.2.1 Pearson’s Chi-square
To test for significant differences between rapists and sexual murderers on offender and
general lifestyle characteristics, relationship and sexual lifestyle characteristics, childhood
victimization, pre-crime factors, victim characteristics, modus operandi characteristics, and
forensic histories, Pearson’s chi-square tests were performed. Pearson’s chi-square tests the
independence of two categorical variables and whether these variables, represented in a
contingency table, differ in expected frequency from each other that would be expected if
they had happened by chance (Field, 2009). There are two assumptions of chi-square tests: 1)
each person falls into only one cell of the contingency table; and 2) the expected frequencies
for each cell should be greater than five. The second assumption of chi-square (that expected
cell frequencies were greater than five) was violated in some instances in Chapter 4
indicating a deviant chi-square sample distribution; therefore, Fisher’s Exact significance was
used where the cell count was violated. Fisher’s Exact computes the exact probability of the
chi-square statistic when sample size is small (Field, 2009).
Phi was used as the measure of the strength of the association between the categorical
variables, where 2x2 contingency tables were generated. Phi restricts the range of the test
statistic between 0 and 1 and is based on a moderated chi-square statistic, taking the sample
size and degrees of freedom into account (Field, 2009).
As many comparisons were made in Chapter 4 between rapists and sexual murderers
a Bonferroni correction 10 (α’= α/k) was applied in order to correct for the potential of an
inflated Type I error, which can happen when performing multiple tests of comparison on a
single dataset (Field, 2009). Therefore, the resulting level of alpha for significance after
correction in Chapter 4 was < 0.001. The drawback to this correction is the loss of statistical
power and the inflation of Type II error, meaning that significant results may be missed
(Field, 2009).
3.2.2 Independent Sample t-Tests
Independent sample t-tests compare two means from independent samples and tests whether
they differ significantly from one another (Field, 2009). Independent t-tests were calculated
for the interval data to test whether the rapists and sexual murderers differed significantly
with regards to their age at their first offence as well as their age at the index offence.
10
Although, when a large number of tests are performed this correction can be too strict (Field, 2009).
48
3.2.3 Mann Whitney U
Mann Whitney U is a non-parametric test that looks at the differences between two
independent samples. Mann Whitney U was calculated as the assumption of a normal
distribution was violated for victim age and “offender age when sexual abuse started” and
therefore, independent t-tests could not be performed. Nonparametric tests are more robust
when the assumptions of parametric tests are violated (Field, 2009). Mann Whitney U tests
were used to explore the difference between the victim age of rapists and sexual murderers as
well as to explore the differences between the age at which rapists and sexual murderers
started experiencing sexual abuse.
3.2.4 MANOVA
Multivariate Analysis of Variance (MANOVA) is an extension of Analysis of Variance

(ANOVA; which is used to compare two or more means for any reliable differences between
them), where main effects and interactions are assessed on a combination of dependent
variables. It tests whether the differences between the mean differences among groups on the
dependent variables is likely to occur by chance. It is designed to look at several dependent
variables simultaneously while controlling for the inflation of familywise error rates (Type 1
error) (Field, 2009). MANOVA also allows for the exploration of the relationship between
the different outcome variables that would not be available by doing multiple ANOVAs on
the individual dependent variables. It therefore has more power than ANOVAs to detect
effects as it takes into account the correlations between dependent variables. MANOVA was
used to examine the relationship of crime type (rape versus sexual murder) and personality
score on the MCMI-III to determine if these two groups differed with regards to their
personality profiles. Eta-squared measures the amount of total variance which is the result of
an effect and which is calculated as the ratio of the effect variance to the total variance
( ƞ
2
=SSeffect/SStotal) was used as the measure of effect size and strength of association. In
general, 0.01 is considered a small effect size, 0.06 medium, and 0.14 a large effect size
(Cohen, 1992).
3.2.5 Hierarchical Cluster Analysis
Cluster analysis is a method of grouping a set of cases or objects together based on their
similarity to other members of the cluster (Field, 2009). The degree of similarity is measured
by either similarity coefficients or dissimilarity coefficients. The correlation coefficient, r,
49
measures the similarity between two variables and in theory could be applied to measure the
similarity between two people to see if their patterns of responses are the same. While the
correlation coefficient is standardized, and is therefore not affected by dispersion differences
across variables, it ignores information about the elevation of scores – it does not tell us the
distance between two people’s profiles (Field, 2009).
The alternative is the Euclidean distance, d, which is the geometric distance between
two objects or cases. The differences between a set of scores are calculated, which can be
both positive and negative, these differences are squared so they are all positive in
denomination, and then added together. Once all the squared differences for all the variables
we are interested in have been added together the square root is taken (to revert back to
original units of measurement), with smaller Euclidean distances being indicative of more
similar cases. The advantage of using Euclidean distances is that is allows for missing data,
however they are also greatly affected by variables with large size or dispersion differences,
so scores need to be standardized (Field, 2009).
Using the similarity coefficients the cases are grouped together. All methods of
cluster analysis begin with all cases being treated as single clusters, and then they are merged
based on a criterion specific to the chosen clustering method. Agglomerative Hierarchical 11
clustering method was chosen as it starts with each case as single clusters, joins together
similar observations, and then repeatedly merges the two closest clusters until a single, all
encompassing cluster is left (Milligan & Cooper, 1987). This is represented using a
dendogram (see Appendix A), which is a visual representation of the distance at which
clusters are combined. When the distances between the sequential vertical lines became
large, indicating increased dissimilarity, this was used as a determination of meaningful
clusters.
Ward’s method was used 12 in part to maximize the differences between clusters, as it
assumes that a cluster is represented by its centroid, and is distinct from other methods as it
uses an analysis of variance approach to evaluate the distances between the clusters and
attempts to minimize the sum of the squared distances of points from their cluster centroids
(Tan, Steinback, & Kumar, 2005). The means for all the variables are calculated and the
11
As opposed to divisive hierarchical clustering, which starts with all objects in a single group and splits them up into smaller groups until
there each their own individual group, or non-hierarchical clustering, such as k-means, which start with a single object and cluster other
objects that are similar to the first one in.
12
As opposed to single linkage, nearest neighbour, (where dissimilarity between two clusters is measured by the minimum dissimilarity
between all combinations of two objects, one from each cluster); complete linkage, furthest neighbour, (where the dissimilarity between two
clusters is measured by the maximum dissimilarity between all combinations of two objects); or average linkage, group average or mean,
(where the dissimilarity between two clusters is measured by the average of all dissimilarities between all combinations of two objects)
(Quinn & Keough, 2001) as they are affected by the structure of the data (level of dissimilarity) (Ludwig & Reynolds, 1988).
50
squared Euclidean distance for each of the cases to the cluster means is calculated and
summed together. When the smallest increase in the overall sum of squared distances is
reached the merging of clusters stops. Ward’s method has been found to be robust and
consistently better for recovering clusters from bivariate data as well as multivariate data (see
Milligan & Cooper, 1987; for a review of validation studies), and thus deemed most
appropriate for the mixed variable type present in the current sample.
The cluster analysis was used to determine the group membership of the sample with
regards to the thematic clusterings found in other studies, such as Sadistic, Angry,
Opportunistic, and/or Compensatory. Each of these has their own descriptive variables that
were used to cluster the cases and describe cluster membership. While agglomerative cluster
analysis may be affected by the way the variables are ordered, and by the removal of cases
(as this can affect the course in which the analysis progresses) (Field, 2009), as well as the
fact that once a cluster is formed it cannot be broken apart later, it was chosen over
Multidimensional scaling (MDS), specifically Smallest Space Analysis (SSA), another
method often used to explore and structure the relationship between variables and determine
groups. The associations between a group of variables is represented as distances in
multidimensional space, with the corresponding distances representing the correlational
relationship between every variable with every other variable. The closeness of fit is between
the distances is carried out by an iterative algorithm and the degree of fit between the original
association and the distances in the space is measured by the coefficient of alienation, often
Jaccard’s coefficient (an asymmetrical measure of behavioural co-occurrences) for
dichotomous data or Pearson’s coefficient for categorical data. The better the fit of the plot to
the data is indicated by the smaller the coefficient of alienation is (Guttmann, 1968).
Problems of replication has plagued the use of MDS in statistical research and highly
correlated variables distort the plots, pulling highly correlated variables into the central area
and pushing less correlated variables to the outer region of the plot (Sturidsson, Langstrom,
Grann, Sjostedt, Asgard, & Aghede, 2006). The issue with this is the fact that the variables
within the centre of the plot are also the highest frequency ones, meaning they tend to co-
occur with all other variables, while the ones on the outside or periphery of the plot are the
lower frequency variables, and form the extreme points of the underlying facet. Combined
this therefore makes reliable interpretation of the facet impossible. Caution also needs to be
taken in the number of dimensions used to obtain the best fit. While two dimensions are
more often used as this allows for better ‘readability’ of the data, this squeezing of data may
result in a poor and highly distorted representation of the data (Jaworska & Chupetlovska-
51
Anastasova, 2009). More than two dimensions and the ability to comprehend become
increasingly difficult, and the structure may be more a product of noise than the essential
structure of the data (Steyvers, 2002).
Discriminant function analysis (DFA), used to determine whether as set of continuous
or binary variables is effective in predicting category membership (Field, 2009), was not used
even though the potential number of groups could have been determined a priori based on
previous research, as Chapter 4 was interested in determining the number of groupings from
the data, and whether they coincided with the previously found groupings (i.e., angry,
sadistic, opportunistic).
3.3 Chapter 5, 6, and 7 Methodology
In Chapters 5, 6, and 7 the relationship between offender characteristics, offence

characteristics, and potential mediators (context, motivations, and implicit theories) are
explored using mediation analysis.
3.3.1Coding of Variables
3.3.1.1 Offender Characteristics
The offender characteristics chosen are those that would typically be found within a profile.
These included: 1) the perpetrator’s age at time of offence; 2) relationship status at time of
offence; 3) whether they have any previous convictions (inclusive of convictions for sexual,
violent, or ‘other’ offences); 4) whether they lived alone or not at the time of the offence; and
5) whether they were employed or not at the time of the offence. Table 3.2 shows the
breakdown of the offender variables and the how they were coded. Perpetrator age was
recorded for on the Functional Analysis questionnaire for 70 of the 102 sample of sexual
aggressors. Relationship status referred to whether the offender was in a committed sexual
relationship at the time of their offence. Whether the offender had any previous convictions
was coded if they had indicated as having any previous sexual, violent, or theft related, or any
other convictions at the time of committing their offence. The offender was coded as either
living alone or as ‘other’, which included living with friends, their parent(s), wife, family
member, or girlfriend, being in the armed forces, living in a hostel, or if they indicated
having no fixed abode.
52
Table 3.2
Offender Characteristics and their Coding
Offender Characteristics (X) Coding
Perpetrator Age Interval; 14 to 57 years
Relationship Status 0=not in a sexual relationship
1=yes in a sexual relationship
Any Previous Convictions 0=no
1=yes
Lives Alone 0=‘other’
1=alone
Employed 0=‘other’
1=employed/student
3.3.1.2 Offence Characteristics
There were six offence characteristics that were focused on for the current thesis; Table 3.3
lists the offence characteristics and their subsequent coding. These were those that would be
identifiable from the crime scene assessment and victim. The level of aggression was based
on the original classification by Feshback (1964), who proposed that aggression could be
‘hostile’ (expressive) or ‘instrumental’. Instrumental violence is the utilisation of violence in
order to achieve a goal, its use is not for the harming of the victim, but as a means to commit
the offence; therefore, aggression occurs if the goal or objective is obstructed (e.g., victim
resistance). Conversely, expressive aggression goes beyond what is necessary in order to
commit the offence, and is utilised in its own right as a means of inflicting harm. The level of
injury inflicted upon the victim ranged from minor (slight damage with/without weapon),
medium (treatment required but no overnight stay) to major (hospitalisation required) 13.
Weapon use was coded as ‘used’ or ‘not used’; not used included a weapon being present but
not physically used against the victim, and a weapon being mentioned but again not being
physically used during the offence. The coding of sexual penetration variables included those
variables, which happen less frequently and could potentially be more pragmatically useful in
differentiating between crime series and offenders (Goodwill & Alison, 2007). Sexual aspects
of the offence that involved fondling or touching of the victim’s body (46%), or the touching
or penetration of the victims’ vagina with either a finger (35%) or penis (64%) was excluded
13
Similar ratings to those used by Quinsey and Chaplin (1982).
53
as these occurred in high frequency and are not pragmatically useful in differentiating
between offenders or offences.
Table 3.3
Offence Characteristics and their Coding
Offence Characteristics (X) Coding
Level of Aggression 0=none/unknown
1=to control victim/instrumental
2=beyond controlling the victim/expressive
99=unknown
Level of Injury 0=none/no injuries
1=minor injuries
2=medium level of injuries
3=major injuries
99=unknown
Offence Outcome 1=murder
2=rape
Victim Age Interval
16 to 86 years
Weapon Used 0=not used
1=used
99=unknown
Sexual Penetration 0=none
1=anal penetration with finger
2=anal penetration with penis
3=foreign object penetration vagina or anal
3.3.1.3 Potential Mediators
There were three different sets of potential mediators explored in Chapter 5, 6, and 7 (see
Table 3.4). The contextual mediators explored in Chapter 5 included the use of drugs
during/or directly before the offence, the use of alcohol during/directly before the offence, the
location of the initial contact between offender and victim, the location of the actual offence,
and the level of victim resistance.
54
For Chapter 6 the presence of the primary Implicit Theories was coded for. While, the
presence of secondary ITs were evident for the current analysis the focus was on the most
prominent IT, which included 1) Women are Unknowable/Dangerous; 2) Women as Sex
Objects; 3) Male Sex Drive is Uncontrollable; 4) Dangerous World; and 5) Entitlement.
Chapter 7 looks at the potential motivations behind the sexual offence based on those
found in previous literature (e.g., Proulx & Beauregard, 2009) – Angry, Sexually
Opportunistic, Sexually Compensatory, and Sadistic.
55
Table 3.4
Potential Mediators for Chapters 5, 6, and 7, and their Coding
Potential Mediators (M): Coding
Contextual Variables
Use of Drugs During the Offence 14 0=none
1=yes
Use of Alcohol During the Offence 0=none
1=yes
Location of Initial Contact between Victim 0=indoor
and Offender 1=outdoor
999=unknown
Location of the Offence 0=indoor
1=outdoor
999=unknown
Victim Resistance 0=nothing/unknown
1=pleading/trying to talk offender out of
offence
2=shouting help
3=verbally hostile towards offender
4=physically hostile towards offender
Implicit Theories Coding
Women Unknowable/Dangerous 0=not indicated
1=indicated
Women as Sex Object 0=not indicated
1=indicated
Male Sex Drive is Uncontrollable 0=not indicated
1=indicated
Dangerous World 0=not indicated
1=indicated
Entitlement 0=not indicated
1=indicated
Motivations Coding
Angry 0=not indicated
1=indicated
Sexual: Opportunistic 0=not indicated
1=indicated
Sexual: Compensatory 0=not indicated
1=indicated
Sadistic 0=not indicated
1=indicated
3.3.2 Mediation Analysis
A regression approach mediation analysis was used to analyse the data in the following
chapters to test the associative link between offence characteristics and offender
14
There was no information available on the actual amount of drugs or alcohol taken; therefore, no indication of the actual level of
intoxication.
56
characteristics taking into consideration contextual variances (Chapter 5), motivational
aspects (Chapter 7), as well as, how the offender views the world (Chapter 6).
Mediating variables and mediation form the basis for many psychological theories
(e.g., social and cognitive psychology), and have applications in prevention and treatment
research (MacKinnon, Fairchild, & Fritz, 2007). It stems from the work done by Lazarsfeld
(1955) and Hyman (1955) on the elaboration of the X  Y relationship in order to better
understand this relation and to test whether this relationship is genuine or spurious. Examples
of its use in psychological research can be found throughout the different areas of
psychology, such as evolutionary psychology, health psychology, and educational
psychological (MacKinnon et al., 2007). For example, mediation analysis could be used to
look at whether attachment style (anxious versus secure) affects the perceived support from
close family members, which may in turn affect the presentation of mental health symptoms?
Mediation analysis looks at the questions: Does the IV directly affect a specific DV or
does the IV affect the DV through an intermediary, or mediating variable (Tabachnick &
Fidell, 2001). Mediation looks at how “an independent variable (X) affects a dependent
variable (Y) through one or more potential intervening variables, or mediators (M)” (Preacher
& Hayes, 2008, p.879). Mediation involving only one mediating variable is called simple
mediation (see Figure 3.1; which is the model of mediation used in the corresponding
chapters). This figure visually represents how an independent variable’s (X) causal effect can
be disseminated into its indirect effect on the dependent variable (Y) through the intervening
variable, or mediator, (M) and X’s direct effect on Y (path c’) (Preacher & Hayes, 2008). Path
a represents the effect or influence of the independent variable (X) on the proposed mediator
(M; predicting M from X). Path b is the effect or influence of the proposed mediator on the
dependent variable (Y) controlling for the effect of X (predicting Y from M). Finally, path c’
is the direct effect or influence of the independent variable (X) on the dependent variable (Y)
that is independent of the effect of M (predicting Y from X) (Hayes, 2009). The total effect of
all variables influences on each other in a mediated model can be expressed as the sum of the
direct (path c’) and indirect effects (product of ab), mathematically expressed as the equation:
c = c’ + ab. Similarly, c’ is the reverse, that is: c’ = c – ab (Preacher & Hayes, 2008). Figure
3.2 shows the total effect of X on Y in an unmediated model.
In the current analyses if Path a, and b are significant, but not path c’, full mediation
can be reported to have occurred, whereas, if path c’ is also significant than it is considered a
partial mediation, as there remains a direct effect of X on Y that is not the result of the
intervening/mediating variable.
57
a M b
X Y
c’
Figure 3.1. Simple Mediation Model. c’ is the direct effect of X; product of a and b quantifies
the indirect effect of X on Y through M.
c
X Y
Figure 3.2. Total Effect of X on Y.
Other methods of mediation analysis (i.e., causal steps approach; Baron & Kenny, 1986)
require Path c (the total effect) to also be significant in order to deem M a mediator, yet if X’s
effect on Y is partly a resultant of an indirect effect through M then this criterion is unlikely to
detect this effect. Furthermore, mediation analysis is an approach to test the intervening
effects of variables, if one accepts the null hypothesis based on a non-significant total effect
then the analysis is failing to test what it was meant to investigate – the intervening effect
between variables. Even if X and Y are not related to one another, it is still possible for M to
be causally related to X and Y (Hayes, 2009). A further criticism of the causal steps approach
is with regards to its low power – if X’s effect on Y is partly through the intervening or
mediating variable M, this approach is unlikely to detect any effect (Hayes, 2009). While this
approach is easy to learn, understand, and use the limitations can result in the researcher
wrongly accepting the null hypothesis, when in fact there is an indirect effect present. The
fact that an indirect effect of X can be exert on Y through M without a significant association
between X and Y, is possible because the total effect is the measure of the sum of both the
direct and indirect effects, which may not all be part of the formal model (Hayes, 2009).
The Sobel Test, the test statistic for testing the null hypothesis that the indirect effect
is zero, is often used as a supplement to the causal steps approach (Hayes, 2009). It is the
product of coefficients approach, and requires an estimate of the standard error of ab, and the
ratio of ab to its standard error (Hayes, 2009; Sobel, 1982, 1986). However, the use of this
test to confirm the validity of the results of the causal steps approach does not provide any
58
further information than running the Sobel test on its own. This test again has its limitations –
it requires that the sampling distribution of the indirect effect is normal (Hayes, 2009), which
in the case of the sample distributions of ab is not the case as these tend to be asymmetrical,
skewed, and liable to kurtosis (Bollen & Stine, 1990; Stone & Sobel, 1990).
3.3.3 Bootstrapping
Bootstrapping, a nonparametric re-sampling procedure which estimates the properties of the

sampling distribution from the sample set (Field, 2009), was performed as a part of the
mediation analysis to determine the indirect effects, ab paths. In this method, the sample is
treated as a population from which smaller samples are repeatedly taken (replacing the data
drawn first before re-sampling; known as Monte Carlo re-sampling). The statistic of interest
(i.e., b coefficient) is calculated for each of the samples drawn, from which the sampling
distribution is estimated. From the standard deviation of the re-sampled distribution a
standard error is estimated, from which confidence intervals and tests of significance can be
calculated (Field, 2009). There are several options for computing confidence intervals: the
normal approximation method 15, the percentile method 16, the bias-corrected (BC) method, the
bias-corrects and accelerated (BCa) method 17, and the approximate bootstrap confidence
(ABC) method 18 (Efron & Tibshirani, 1986).
With regards to mediation, bootstrapping represents the sampling distribution of the
indirect effects as it treats the obtained sample size n as a representation of the population,
with the researcher determining the total number of times, k, that it is re-sampled (Hayes,
2009; Preacher & Hayes, 2008). The argument being that it is better to make inferences based
on the sample at hand than to make potentially non-viable ones about the population as a
whole (Mooney & Duval, 1993).
By repeatedly re-sampling k number of times, often in the thousands, empirical
approximations of the sampling distribution of the indirect effect (ab) can be developed. This
is then used to construct confidence intervals for the indirect effect by sorting the k values of
ab from smallest to largest. In order to reject the null, that the indirect effect is zero, zero
15
Computes the approximate standard error using the sampling distribution from all of the bootstrap re-samples and uses the z-distribution
to compute the confidence intervals (Haukoos & Lewis, 2005).
16
Uses the frequency histogram of the m statistics (number of bootstrap samples) which are computed from the bootstrap samples and sets
the limits of the 95% confidence intervals at the 2.5 and 97.5 percentiles (Haukoos & Lewis, 2005).
17
Is an adjustment of the percentile method that adjusts for the skewness and non-constant variances in the bootstrap sampling distribution
(Haukoos & Lewis, 2005).
18
An approximation of the BCa method, which requires a smaller number of re-sampled data, sets (Haukoos & Lewis, 2005).
59
must not be contained between the lower and upper bound of the intervals (Hayes, 2009;
Preacher & Hayes, 2008). The advantages of bootstrapping are that it maintains reasonable
control over Type I errors and has a high level of power (MacKinnon, Lockwood, Hoffman,
West, & Sheets, 2002; MacKinnon, Lockwood, & Williams, 2004). In addition,
bootstrapping makes no assumptions about the shape of the sampling distribution (unlike the
Sobel test) as the inferences made from bootstrapping are based on an estimate of the indirect
effect itself (Hayes, 2009).
A disadvantage of bootstrapping is that it can be computationally intensive, although
this is often offset by the speed of current computers and computer programmes. Another
limitation is that the re-sampling is based on the sample in hand, meaning that if the sample at
hand is not representative of the population; there is the possibility that the re-sampled
sample still may be unrepresentative of the population. This would limit the ability to
generalise outside of the current sample. In other words, the original sample is assumed to
reflect the variety and range of the population, and if this is not the case, then the random
sampling performed by bootstrapping results in more sampling error and invalid statistical
estimations (Haukoos & Lewis, 2005).
3.3.4 Preacher and Hayes Estimate of the Indirect Effects
A macro developed by Preacher and Hayes (2008) 19 called ‘Indirect’ was installed and run as
a custom dialog in the regression drop down menu in PASW Statistics 18. The Indirect script
is designed to estimate the total, direct, and single-step indirect effects of causal or
independent variable X on an outcome or dependent variable Y through a proposed mediator
variable M (or a list of mediator variables). It calculates the Sobel test for both the total and
specific indirect effects, as well as, calculating the percentile-based, bias-corrected, and bias-
corrected and accelerated bootstrap confidence intervals for the indirect effects. The macro
automatically detects whether the outcome variable is continuous or binary and estimates
accordingly. If binary outcome is detected the direct and total effects, and the path(s) from
the proposed mediator(s) are estimated using logistic regression. Otherwise, the estimates of
the paths are calculated using Ordinary Least Squares regression.
19
Macro can be found at: http://www.afhayes.com/spss-sas-and-mplus-macros-and-code.html
60
3.3.5 Assumptions of the Mediation Analysis
3.3.5.1Multicollinearity
Multicollinearity occurs when the independent variables are too highly correlated
(Tabachnick & Fidell, 2001). It only poses a problem in multiple regression since this
analysis requires more than one predictor, which is also a problem for logistic regression
(Field, 2009). One indication of multicollinearity, besides inspecting the correlation matrix
for correlations above 0.80, is inflated error terms. Large standard errors indicate that the b
coefficient for the sample is more variable, and is less likely to represent the population
(Field, 2009). Multicollinearity can also affect the width of the resulting confidence
intervals, resulting in very wide intervals. Wide confidence intervals are highly affected by
the exclusion or addition of a data point, which can change the coefficients drastically
(Myers, 1990). The issues with highly correlated independent variables is the fact that they
explaining the same part of the variation in the dependent variable. This makes it hard to
determine which variable is responsible for the variance, and lessens their explanatory power,
as well, the significance of their coefficients are "divided up" between them (Myers, 1990).
Issues of collinearity can also mask the importance of the individual predictors – if the
variables are highly correlated and each accounts for similar variance with regards to the
outcome, it is unclear which variable is more important within the model as a whole.
The variance of inflation (VIF) and the tolerance (T) statistic are two measures of
collinearity. The VIF indicates whether one predictor has a strong linear relationship with the
other predictor(s), and a value of 10 is thought to be a good value to use as a cut-off.
Although, a value of 1 or more can often indicate that multicollinearity may be influencing
the regression model (Bowerman & O’Connell, 1990). Related to the VIF is the measure of
tolerance (T), which is the 1/VIF, where values below 0.1 indicate severe problems of
collinearity, although values below 0.2 should be taken note of (Fields, 2009). For each of the
chapters, multicollinearity was assessed using the VIF and T values (these are reported in the
individual chapters).
61
3.3.5.2Linear Regression and Logistic Regression Assumptions
The main assumptions of linear regression are (Field, 2009):
1) Normal distribution of errors; this assumes that the differences between the residuals
of the model and the actual observed data are zero or close to zero.
2) Linear relationship between the dependent and independent variables – “the mean
values of the outcome variable for each increment of the predictor(s) lie along a
straight line” (Field, 2009, p. 221); if violated the analysis will under-estimate the true
relationship and limit the generalisability of the findings.
3) Independence, which is the assumption “that all of the values of the outcome variables
are independent...[and] from a separate entity (Field, 2009, p. 221);
4) Homoscedasticity is the premise “that at the residuals at each level of the predictor(s)
should have the same variance” (Field, 2009, p. 220); when violated, the occurrence
of marked heteroscedasticity, it can lead to serious distortion of findings and increase
the possibility of Type 1 error (Tabachnick & Fidell, 2001).
5) Multicollinearity
If any of these assumptions are violated then the predictions yielded by the regression model
could be seriously biased and misleading meaning that generalisations cannot be made
beyond the current sample.
Logistic regression shares some of the same assumptions of linear regression, namely
linearity, independence, and multicollinearity.
3.3.6 Measurement of Variance
As part of the output by the macro different measurements of variance, and how well the
model fits the data, are generated depending on whether linear regression or logistic
regression is run and used to describe the amount of variance explained by the models for the
different mediation analyses run.
3.3.6.1 R-square Statistic
The R-square statistic is a measure of how well a regression line in linear regression
approximates the real data points, and indicates how successful the fit of the model is in
explaining the variation of the data. It is the proportion of variance shared by the outcome
62
and predictor variables. R-square can take any value between 0 and 1, with values closer to 1
indicating that a greater proportion of variance is accounted for by the model (Field, 2009).
3.3.6.2 Wald Statistic

The Wald statistic, based on a chi-squared distribution 20, indicates whether the b coefficient
in logistic regression models 21 for the predictor is significantly different from zero (Fields,
2009). It is the b coefficient divided by the standard error, and specifies if the independent
variable is a significant predictor of the outcome. However, when the coefficient is large this
can result in an inflated standard error leading to the Wald statistic being underestimated. The
result is a possible increase in Type II error (accepting the null) (Fields, 2009).
3.3.6.3 Cox and Snell R Square and Nagelkerke R Square
The Cox and Snell R2 statistic is the logistic regression model version of the coefficient of
determination, which “is based on the log-likelihood of a model (LL(new)) and the log-
likelihood of the original model (LL(baseline)), with a sample size n” (Field, 2009, p. 784)
Although, it does not reach its maximum value of 1 (Field, 2009). Nagelkerke’s R2 is another
form of the coefficient of determination for logistic regression and a variation of the Cox
Snell’s R2, which overcomes the limitation of the Cox Snell R2 by being able to reach its
maximum value of 1 (Field, 2009). Both of these are analogous to Pearson’s correlation
coefficient, R2, in linear regression, and describe the proportion of variance in one variable
that is explained by a second variable (Field, 2009).
20
A probability distribution of the sum of squares of the several normally distributed variables (Field, 2009).
21
It is the equivalent to the t-statistic in linear regression (Field, 2009).
63
Chapter 4
Rapists and Sexual Murderers: Combined Pathways to Offending
The aim of this is to determine if there are sufficient differences or similarities to warrant
combing the current sample of sexual murderers and rapists into one sample of sexual
aggressors of adult women. The idea of sexual offending being an escalating continuum is
discussed. The pathways to offending of sexual aggressors of women are explored
concerning the offender’s general and sexual life-style, their own victimization, previous
criminal history, pre-crime factors, as well as the victimology, and modus operandi. These
factors are discussed with regards to the Angry, Sadistic, and Compensatory pathways found
in previous research.
The following chapter was accepted for publication in the book: Pathways to Sexual
Offending, due for publication summer 2012. The authorship on the chapter indicates
collaborative working. To clarify, I am the senior author and my supervisors Anthony Beech
and Jessica Woodhams are also named as authors.
64
Chapter 4: Rapists and Sexual Murderers: Combined Pathways to Offending
4. Introduction
The aim of this chapter is to compare rapists and sexual murderers on a set of variables
encompassing developmental, psychological, criminological, and offence-related factors. It
will be determined whether they should be considered as two separate groups of offenders or
as representing individuals who display behaviour that forms a continuum of sexual violence.
Secondly, the chapter will investigate the specific pathways to offending in a UK sample of
sexual aggressors against adult females.
4.1 Comparing Rapists and Sexual Murderers
In general terms, the literature on sexual offenders tends to put offenders into types and
examine them in isolation; this literature concentrates on paraphilias (e.g., Abel & Osborn,
1992; Bradford, Boulet, & Pawlak, 1992; Kafka, 1997); child sexual abuse (e.g., Mian,
Wehrspann, Klajner-Diamond, LeBaron, & Winder, 1986; Mrazek, Lynch, & Bentovim,
1983; Proulx, Perreault, & Ouimet, 1999; Sheldon, & Howitt, 2008); rapists (e.g., Canter,
Bennell, Alison, & Reddy, 2003; Canter & Heritage, 1990; Groth, 1979; Knight, 1999;
Kocsis, Cooksey, & Irwin, 2002; Langton & Marshall, 2001); non-serial sexual murderers
(e.g., Brittain, 1970; Burgess, Hartman, Ressler, Douglas & McCormack, 1986; Fisher &
Beech, 2007; Kraemer, Lord, & Heilbrun, 2004; Meloy, 2000; Porter, Woodworth, Earle,
Drugge & Boer, 2003; Ressler, Burgess & Douglas, 1988); sadistic sexual offenders (e.g.,
Brittian, 1970; Dietz, Hazelwood, & Warren, 1990; MacCulloch, Snowden, Wood, & Mills,
1983; Marshall & Kennedy, 2003;); and serial sexual murderers (e.g., Douglas, Burgess,
Burgess, & Ressler, 1992; Egger, 1984; Ferguson, White, Cherry, Lorenz, & Bhimani, 2003;
Silva, Leong, & Ferrari, 2004; Warren, Hazelwood, & Dietz, 1996). Although, these studies
provide valuable information about these different types of sex offenders in isolation they do
not consider or test whether these types of sex offender are truly different from one another
With regards to sexual aggressors of adult women, few studies have directly
compared rapists 22 and sexual murderers 23. Those that have compared these two groups,
22
“Rape: A person (A) commits an offence if a) He intentionally penetrates the vagina, anus or mouth of another person (B) with his penis,
b) B does not consent to the penetration, and c) A does not reasonably believe that B consents.” (Sexual Offence Act 2003)
23
“The term sexual murderer is applied to individuals who have killed someone where there is either clear forensic evidence of a sexual
element to the killing, or a sexual component is admitted or suspected” (Beech, Oliver, Fisher & Beckett, 2005, p. 23).
65
generally concentrated on their developmental, psychological, and criminological differences,
and the modus operandi of the sexual offenders. Developmentally, it was found that sexual
murderers, in contrast to rapists, were more frequently socially isolated in both childhood and
adolescence, indicating a general lack of friends, feelings of being isolated or excluded from
their peer group, and a lack of sexual relationships (Grubin, 1994; Milsom, Beech & Webster,
2003; Oliver, Beech, Fisher, & Beckett, 2007; Proulx, Beauregard, Cusson, & Nicole, 2007).
The sexual murderers more often came from families with violent fathers (Langevin, Ben-
Aron, Wright, Marches, & Handy, 1988; Proulx, et al., 2007) and were the victims of both
sexual and physical childhood abuse (Milsom, et al., 2003; Proulx, et al., 2007).
Psychologically, the sexual murderers showed higher rates of Antisocial Personality
Disorder 24 (Langevin, 2003; Langevin, et al., 1988; Proulx, et al., 2007), although this was
found to be rare amongst both groups (Oliver, et al., 2007; Proulx, et al., 2007). They were
also found to have a higher prevalence of paraphilias, such as sadism and transvestism, than
rapists (Grubin, 1994; Langevin, 2003; Langevin, et al, 1988; Proulx, et al., 2007).
While both the rapists and sexual murderers tended to have varied histories of
offending, the sexual murderers started their offending career at an earlier age than the rapists
(in their early to mid teens) (Langevin, 2003; Proulx, et al., 2007; Oliver, et al., 2007), and
had committed their first sexual murder by approximately 20 years of age (Langevin, 2003).
The offences of the sexual murderers were more frequently preceded by feelings of anger
(Grubin, 1994; Langevin, et al., 1988; Milsom, et al., 2003; Proulx, et al., 2007), and the use
of alcohol (Grubin, 1994; Langevin, 2003; Proulx, et al., 2007), although the sexual
murderers were not necessarily intoxicated. As well, they tended to commit their assault
against female strangers (Langevin, 2003; Langevin, et al., 1988; Proulx, et al., 2007), whom
they strangled (Grubin, 1994; Langevin, et al., 1988).
The sample sizes of these comparison studies are small (i.e., Langevin et al., 1988;
Milsom et al., 2003), and only a narrow set of variables was investigated (Grubin, 1994;
Milsom et al., 2003), thus restricting generalizability and the possibility for comparisons
(Proulx, Cusson & Beauregard, 2007). However, on the whole, there remain a limited number
of differences between sexual murderers and rapists, with more similarities apparent within
their backgrounds, criminal histories, and personalities.
24
The presence of personality or clinical disorders was measured using different tools in the different studies. Millon Clinical Multi-axial
Inventory III (Million, 1994) was used in Langevin, et al. (1988), Oliver, et al. (2007), Proulx, et al. (2007). Antisocial Personality
Questionnaire (Blackburn & Fawcett, 1996, 1999) was also used in Oliver, et al. (2007).
66
4.1.1 Escalation and continuum of sexual violence
These few comparative studies show that sexual murderers have only been distinguished
from other sexual offenders on a small number of variables; most of which are static
background characteristics. These results support the idea that sexual homicide and rape
represent two ends of a single continuum of sexual violence (Salfati & Taylor, 2006). For
example, there appear to be similar motivations across the two groups, with sexual murderers
acting out more extreme forms of the motivations identified in rapists (Beech et al., 2005;
Oliver et al., 2007), which ultimately lead to the death of their victim. Studies looking at the
escalation of sexual offending have found that even at the ‘lower’ end of the sexual offence
continuum (i.e., exhibiting and obscene phone calls) there is evidence of escalation for some
sexual offenders towards more serious and physically violent sexual offences (i.e., sexual
assault) (Stermac & Hall, 1989). As well, it has been found that convicted sexual offenders
are seven times more likely to be convicted of a subsequent homicide than the general
population, even though the likelihood of a sex offender going on to commit a subsequent
homicide is still low (1 in 400) (Francis & Soothill, 2000).
In sexual murder and rape, the offender-victim interactions are primarily set apart by
the level of violence involved. Rape behaviours are associated toward one end of the
continuum (i.e., blindfolded, binding, clothing ripped, weapon brought to the scene), with
sexual murder behaviours located toward the other end of the continuum (i.e., object
insertion, anal penetration, non-controlled violence, multiple wounding). Those behaviours
positioned in the middle of the continuum (i.e., steal, vaginal penetration, naked, forensic
awareness) represent both forms of sexual aggression (Salfati & Taylor, 2006). However, to
wholly understand sexual violence, it is not enough to continue exploring facets of behaviour
based solely around individual subgroups (e.g., rape or sexual murder), but a move towards
an intra-domain model that tests all the facets of behaviour relating to sexually violent
behaviour is necessary (Salfati & Taylor, 2006). This suggestion is supported by Ch ѐnѐ and
Cusson (2007), who found pre-crime alcohol consumption, presence of a familial, or intimate
relationship, with the victim, and the use of a blunt object as a weapon to have a strong
association with the severity of the crime (e.g., use of excessive force, extent of victim
injury), rather than its outcome. This does not support a diametrical separation of sexual
murderers from rapists. Instead, the differences between them are an indication of the
predictors of the escalation from sexual aggression to sexual murder (Nicole & Proulx, 2007)
67
along a continuum of sexual violence. Related to the continuum of sexual violence are the
pathways to offending of these sexually aggressive offenders.
4.2 Pathways of Sexual Aggression Against Women
Early models of the offence process and pathways, such as Pithers, Marques, Gibat and
Marlatt’s (1983) Relapse Prevention model 25 and Polaschek, Hudson, Ward, and Siegert’s
(2001) Rape Model 26, have been criticised for proposing a single pathway to offending, for
their descriptive nature, their focus on negative affect and the small samples and limited
variables from which they were devised (Proulx & Beauregard, 2002; 2009 a, b). In
response, Proulx and Beauregard (2002; 2009a, b) collected a breadth of information relating
to pre-crime affect and behaviour, modus operandi, victim and offender characteristics and
situational variables from interviews with Canadian incarcerated sexual offenders. These
variables were subject to multiple correspondence analysis 27 and cluster analysis
techniques 28, which identified several distinct pathways to sex offending. In their studies on
the offending process of non-serial Canadian sexual murderers and extra-familial rapists,
Proulx and Beauregard (2002, 2009a) identified a sadistic, angry, and sexually opportunistic
pathway to offending, which differed on the offender’s use of deviant sexual fantasy, level of
planning, use of physical violence, treatment of victim, and victim selection. The offenders in
the sadistic pathway during the hours preceding their crime often engaged in and used
deviant sexual fantasies; they selected, kidnapped, restrained, humiliated and mutilated their
stranger victims, using more force than was necessary to complete the assault. They spent a
lot time with their victims (more than 30 minutes) and the attack often resulted in the death of
the victim, following which they hid the body. Such characteristics have been reported in
studies of sadistic serial murderers (Gratzer & Bradford, 1995; Warren et al., 1996).
In contrast, the murderers in the anger pathway did not plan their offences, preselect
their victim, or use physical restraints. Although, they did not explicitly plan their offence
25
Pithers, Marques, Gibat and Marlatt’s (1983) relapse prevention (RP) model for sexual aggressors was derived from a method of
enhancing maintenance of change in substance abusers originally described by Marlatt and colleagues (Chaney, O’Leary, & Marlatt, 1978;
Marlatt, 1982; Marlatt & Gordon, 1980, 1985) and is based on the idea of a cognitive-behavioural chain. It proposes a variety of factors that
interact with each other, and which influence whether or not a sexual offender will reoffend.
26
Polaschek, Hudson, Ward and Siegert (2001) rape model (RM) of offence processes followed five phases, which the offenders went
through in committing their offences: 1) offender background factors; 2) goal attainment; 3) the approach; 4) the preparation; and 5) the
offence. In sum, Polaschek et al.’s RM is concerned with the explicit goals of the offender and his decision-making throughout the entire
offence process. It focuses on the interaction between the offender and the victim, which is often overlooked in the literature.
27
Multiple correspondence analyses (MCA) provide a visual representation of the associations between multi-level categorical variables
along with descriptive statistics that indicate the number of dimensions of the associations (Clausen, 1998).
28
Cluster analyses are mathematical methods used to determine clusters of similar objects in a set (Romesburg, 2004).
68
they often chose prostitutes as their victims, who they would seriously injury in the course of
the offence. They did not mutilate or humiliate their victim, and they often left the body of
their victim at the crime scene. They reported experiencing anger and being intoxicated just
prior to their crime (while those in the sadistic pathway reported more positive affect before
their crimes), and they did not present deviant sexual fantasies. Finally, the sexually
opportunistic pathway offenders used alcohol preceding their crime and often used minimal
force or violence during their sexual attack on a victim, who was typically an acquaintance.
They did not humiliate, mutilate or injure their victim (Proulx & Beauregard, 2009a).
Extending the results of their previous studies, Proulx and Beauregard (2009b)
investigated pathways to offending based on the pre-crime factors, modus operandi,
situational variables, and adding sexual and general life-style factors, personality disorder
profiles factors associated with the sadistic, angry, and opportunistic processes. The
prominent features of the sadistic aggressors were general lifestyle inactivity, deviant sexual
fantasies and consumption of pornography, generalized conflict with women in the year
preceding their offence, and social isolation. The angry aggressors experienced anger and
generalised conflict with women leading up to their offence, and with prostitutes as the most
likely victims of this group of offenders. The third group of aggressors, the sexually
opportunistic, had generalised conflict with society and women prior to their offence, they
used drugs and alcohol, and had a large number of sexual partners and explicitly planned
their offence.
Another examination of the offence pathways of sexual aggressors that takes into
consideration surface level cognitive distortions and underlying implicit theories, is that of
Beech, Fisher, and Ward (2005), and Beech, Ward, and Fisher (2006). Implicit theories are
causal theories that generate thoughts, feelings, and motivational beliefs (Mann & Beech,
2003) that function to explain other people’s actions and help make predictions about the
world (Polaschek & Ward, 2002). Polaschek and Ward (2002), in their examination of the
motivational beliefs literature, identified five offence-related implicit theories (ITs) that
support and sustain a rapist’s pro-offending attitudes:
1) Women as sexual objects (WSO). Women are seen to constantly desire sex and are
in a constant state of sexual reception, they are seen to exist to meet the sexual
needs of men, even if it is coerced or violent.
69
2) Males’ sex drive is uncontrollable (SDU). Men’s sexual energy is difficult to
control, and women play a key role in the loss of this control by denying
reasonable sexual access.
3) Dangerous world (DW). The world is a dangerous place and it is necessary for the
offender to fight back and achieve dominance and control over other people.
4) Women are unknowable/dangerous (WUD). Women are deceptive in
communicating their desires and needs to men, and are out to trick or con men.
5) Entitlement (E). Men are superior to and more important than women, and have
the right to assert their needs above/over them.
It was found that Danger World and Women as sexual objects were the two most dominant
ITs present in rapists. Consequently, investigating the absence or presence of DW and WSO
ITs, Beech, et al. (2006) identified three main groups in their sample of rapists; namely,
Group 1: violently motivated (presence of DW and absence of WSO); Group 2: sexually
motivated (presence of WSO and absence of DW); and Group 3: sadistically motivated
(presences of DW and WSO). The violently motivated Group 1 offenders were more likely to
attack an (ex) partner, were more likely to have recently split from a sexual partner, and to
have committed previous sexually violent and/or non-sexually violent offences compared to
Group 2. E and WUD ITs were also present in 15-20% of this group. Group 2, the sexually
motivated group, were more likely to offend against younger stranger victims. E ITs was
present in almost three-quarters of this group, suggesting it was another major motivating
factor for their offending. The third group, the sadistically motivated, had sexual and/or
violent offence histories, and were twice as likely as the other two groups to have a known
history of psychiatric problems (e.g., depression, personality disorder). Their motivations for
offending were to carry out their sadistic fantasies, the need for power and domination over
women, and their desire to sexually humiliate women.
Beech, et al. (2005) looked at the presence or absence of these five ITs in a sample of
sexual murderers. Again, three groups were identified, albeit by the absence or presence of
Dangerous World or Males’ sex drive is uncontrollable (the two dominating ITs), Group 1:
grievance and/or anger motivation (presence of DW, absence of any sexually motivating
ITs); Group 2: sexual motivation (dominance of SDU); and Group 3: sadistic motivations
(presence of DW and SDU). The first group motivated by grievance and anger toward
women, where more likely to know their victim, whom they were more likely to beat or stab
to death. This group of offenders were the least likely to sexually mutilate their victim after
70
death and they had the highest level of criminal histories for nonsexual and/or nonviolent
offences. The sexually motivated group (Group 2), targeted adult women, had pre-offence
thoughts and fantasies dominated by the prospect of sexual gratification and having sex.
These offenders often murdered to keep the victim as a means of keeping them quiet during
the offence or to avoid detection afterwards. The third group, the sadistically motivated, held
prior intentions to kill and mutilate their targeted stranger victims, and reported that their
motivation to offend was to act out their violent and sadistic thoughts and fantasies. They had
a reported history of violence against women that was significantly higher than both the other
two groups combined. For some of the sexual murderers, Beech, et al. (2006) hypothesise
that these are extreme forms of the motivations identified in their aforementioned study using
rapists. They argue that it is imperative to continue to look at motivations and the risk of
carrying out further offences because in some cases these offences may escalate from rape to
murder.
The strength of the offence process and pathways to offending models are their basis
in both qualitative and quantitative methods, and their use of a wide variety of variables
concerning the offender, the offence, and the victim (Proulx & Beauregard, 2009b). Sexual
aggressors vary in their primary goals (e.g., sexual gratification versus grievance), their
capacity to plan their offence (e.g., explicit versus implicit), the negative or positive emotions
they experience throughout their offence process (Ward & Hudson, 1998) and the implicit
theories underlying their offences (Beech, et al., 2005; Beech, et al., 2006). The implications
of these models have a bearing on treatment and the identification of specific interventions
for sexual aggressors (Ward, Hudson & Keenan, 1998) 29.
4.3 Methods
4.3.1 Participants
One hundred and two sexual aggressors who had committed and were convicted of at least
one sexual offence against an adult (16 years of age and above) female victim were included
in this study. Sixty-four of the offenders had committed rape as their index offence, while the
remaining 38 were sexual murderers 30. The participants were all convicted sex offenders
taking part in the Core Sex Offender Treatment Program (Core SOTP) run by the UK Prison
29
See Beech, et al. (2006) for a discussion of the treatment implications of the identified sexual offending motivations.
30
For the current study, the term for rapist will be in line with the Sexual Offence Act 2003 and the term sexual murderer will be the same
as those used by Beech, et al. (2005).
71
Service between 1998 and 2002. Based on a ‘cognitive-behavioural’ approach, the Core
SOTP is an intervention programme aimed at increasing offenders’ motivation to avoid re-
offending and developing the self-management skills necessary to accomplish this. It targets
a number of different areas which are considered important in the aetiology and maintenance
of sexual offending behaviour (i.e., deviant arousal; distorted thinking patterns; lack of
empathy; denial and minimisation; patterns of offending). The average length of the
programme is 180 hours, completed in approximately two-hour content sessions, meeting for
two to five sessions per week (Beech, Oliver, Fisher & Beckett, 2005). The sample was
selected from 55 different Core SOTP groups running at seven Her Majesty’s Prison
establishments in England and Wales. This sample represented 2 % of the total male
population under immediate custodial sentence for sexual offences at that time and about
10% of the total male prison population under immediate custodial sentence (Home Office,
2004). The samples’ ages ranged from 14-57 years old at the time of the offence (M = 26.70,
SD=8.60). The mean sentence length for the total sample was 10.70 years (SD=4.35), ranging
from 5 to 25 years.
4.3.2 Procedure
Individual offenders undergoing the Core SOTP were approached and those who consented
to be interviewed were seen for approximately 75 minutes. During this time the Functional
Analysis Interview (FAI), a standard pre-assessment screening interview for the Core SOTP
treatment program, was completed. Their demographic and offence details were obtained
from their official prison file. The FAI consists of 18 questions exploring the cognitions,
emotions, and behaviours of the offenders: 1) prior to the index offence (e.g., major life
events experienced prior to offence); 2) during the index offence; and 3) post index offence.
In addition to being interviewed, 64 of the 102 offenders completed the Millon
Clinical Multiaxial Inventory-III (MCMI-III; Millon, 1994). The MCMI-III is a standardised,
175-item, self-report personality questionnaire comprised of 14 Personality Disorder scales. It
provides information on specific disorders outlined in the Diagnostic and Statistical Manual
of Mental Disorders IV-TR (DSM-IV-TR American Psychiatric Association, 2000) in terms
of Axis I (clinical disorders) and Axis II (personality disorders). Raw scores are obtained for
11 personality disorders: schizoid, avoidant, dependent, histrionic, narcissistic, antisocial,
obsessive-compulsive, passive-aggressive, schizotypal, borderline and paranoid. These scores
are then transformed into base-rate (BR) scores, with reference to the prevalence of each
72
personality disorder in clinical populations. Finally, the BR scores are compared to cut-off
scores to indicate the level of pathology of the individual. A BR score of 75 or greater
suggests problematic trait features and is suggestive of the presence of disorder, while scores
of 85 or more provide substantial support for the presence and prominence of a personality
disorder or clinical syndrome.
4.4 Results
4.4.1 Findings of the comparisons between the rapist and sexual murderer samples
Prior to investigating the offenders’ pathways to offending, it was first necessary to determine
whether the rapists and sexual murderers were sufficiently similar in their characteristics that
they could be grouped together, or if they differed from one another and therefore should be
analysed separately. The source of information regarding the offender’s characteristics was
the FAI. These data were coded in terms of the presence or absence of characteristics and
thus represented nominal data whereas characteristics such as offender age at first offence
represented interval data. The MCMI also produced interval level data. To compare and
contrast rapists and sexual murderers Pearson’s Chi Square tests were calculated for the
nominal data and Independent Samples t-Tests, Mann Whitney U and MANOVA tests for the
interval data. Fisher’s Exact significance was used where the cell count was violated in the
Chi Square tests. A Bonferroni correction (α new= [1-(1-α) df]/n) was applied correcting for an
inflated Type I error, which can happen when performing multiple tests of comparison on a
single dataset. Therefore, the level of alpha for significance was <0.001 (See Appendix C
Tables B1 and B2).
4.4.1.1Offender and general life-style characteristics
The rapists’ ages ranged from 14 to 57 years and they were older (M = 29.36, SD=9.14),
although not significantly, than the sexual murderers (16-48 years; M = 23.88, SD=7.08).
Sexual murderers were more likely to be White (92%), while a greater proportion of the
rapists were of an ethnic minority (19.6%). Level of employment and unemployment did not
differ between the sexual murderers and rapists, with two-thirds of both groups being in
employment or in education (61%; 60% respectively) at the time of the offence. Just over
20% of the sexual murderers and less than 20% of the rapists lived alone. The majority of the
sexual murderers lived either with their parents (29%) or with a girlfriend or wife (21%),
73
although this did not differ significantly from the rapists with just over one-third of the rapists
also living with their parents or a current girlfriend or wife.
The majority of both the sexual murderers and the rapists had experienced some kind
of major life event just prior to the offence (82%; 86%, respectively) (i.e., relationship break-
up; moving house). However, the sexual murderers more often reported experiencing inter-
relational difficulties prior to the offence than the rapists (21%; 5%, respectively).
4.4.1.2 Relationship and sexual lifestyle characteristics
Rapists were more likely to be in a sexual relationship at the time of the offence (73%),
lasting more than a year in length (52%), and with which they were currently dissatisfied
(64%). The majority of both the sexual murderers and the rapists were currently in a
relationship with a female partner (55%; 72%, respectively); however, sexual murderers were
more likely than the rapists to have male partners (45%; 28%, respectively). On the other
hand, it was common in both groups for the previous partner to have been male (63% of
sexual murderers; 77% of rapists). Neither group reported regularly using the services of
prostitutes. A greater number of the rapists (70%) reported having had long-term
relationships as compared to the sexual murderers (47%), who tended to have had both a few
short-term and a few long-term relationships (47%). The sexual murderers more often
reported having difficulty in making close friends and in keeping friends.
4.4.1.3 Childhood victimization
More of the rapists reported suffering sexual abuse as a child (21%) compared to the sexual
murderers (9%). Although not significant, a higher proportion of the sexual murderers (60%)
reported that a parent was the perpetrator of the sexual abuse while the rapists were more
likely to have been sexual abused by a stranger, friend or another family member (86%). The
median age the abuse started for sexual murderers and rapists was the same, 12 years (Range:
4-15 years). The rapists suffered more contact sexual abuse (86%; i.e., oral sex, intercourse)
than the sexual murderers who were more often exposed to visual sexual abuse (67%; i.e.,
witnessing sex, being shown pornography), although this association was not significant.
The rapists more often reported suffering physical abuse as a child than the sexual
murderers (28%; 7%, respectively). The physical abuse of the rapists often took the form of
74
being smacked or hit (21%), and extreme acts (29%), while the sexual murderers were more
often hit with an object (43%), punched and/or kicked (29%).
4.4.1.4 Age at first offence and previous offences
The sexual murderers had committed their first offence at a younger age (M=22.03,
SD=7.38) than the rapists (M=25.34, SD=8.62), although this difference was not significant.
The majority of offenders in both samples had some kind of history of previous offending
(rapists 77%; sexual murderers 76%). A greater percentage of rapists reported having
committed previous sexual offences (44%), previous violent offences (44%) and ‘other’
previous offences (67%) than the sexual murderers (29%; 40%; 63% respectively). However,
these differences were not significant.
4.4.1.5 Pre-crime Factors
Prior to the index offence, the sexual murderers more often reported feeling under pressure
from their families (53%) and from their current relationship (58%) when compared to the
rapists (22%; 38%, respectively). The majority of both the rapists and the sexual murderer
sample reported they did not feel in control of their life prior to the offence (70%; 76%
respectively). Similarly, feelings of anxiousness and anger were reported by both the rapists
(67%; 58%), and the sexual murderers (74%; 71%), just prior to the offence. Sexual
murderers (45%) more often reported feeling that they had been humiliated prior to the
offence than the rapists (25%). More than 40% of the rapists and approximately one- third of
the sexual murderers were under the influence of some type of drug at the time of the offence,
cocaine being the drug of choice for the rapists. More of the rapists (61%) reported
fantasising about the intended offence within 48 hours of its commission than the sexual
murderers (34%).
4.4.1.6 Victim characteristics
The victims ranged in age from 16 years old to 86 years of age (Mdn=27.00). The rapists
offended against significantly younger victims (Mdn= 25.00; 16-78 years old) than the sexual
murderers (Mdn =35.00; 16-86 years old) (U=864.50, z = -2.44, p < .05, r = -0.24). In
approximately half of the offences, the victims were strangers to the offenders (rapists 61%;
sexual murderers 47%). The most common victim behaviours during the offence were for the
75
victim to try to dissuade the rapist from committing the offence (48%) or to comply with the
rapist’s demands (63%). The frequency of these behaviours in the rapist sample was higher
than in the sexual murderer sample (16%; 32%, respectively). Crying and/or pleading with
the offender were also common victim reactions during the sexual offence (rapists 31%;
sexual murderers 29%).
4.4.1.7 Modus operandi
While evidence of planning could be found in a proportion of both the rapists’ (53%) and the
sexual murderers’ (29%) offences, significantly more of the rapists planned 1) their offence,
2) who their intended target would be, 3) and when the offence would occur. The sexual
murderers typically used a weapon they had found at the scene, which again indicates a more
spontaneous offence. The rapists were more likely to use non-lethal coercion, such as verbal
threats (X2(1) =15.04, p<0.001), threatening the victim with a weapon, and using force with no
serious resulting injury to the victim (X2(1) = 25.83, p <0.001). Although the rapists used
coercion more often, the sexual murderers’ use of coercion resulted in serious injury
significantly more frequently (X2(1) =16.88, p<0.001).
The victims of the sexual murderers were more likely to have been restrained, or
unconscious (X2(1) = 13.20, p < 0.001) during the offence, and although not significant they
were more likely to have been strangled. The rapists used their penis to touch or penetrate the
victim’s vagina more often than the sexual murderers (X2(1) = 22.82, p < 0.001), although,
there were similar levels of occurrence between the groups for the penetration or touching of
the vagina with a finger (sexual murderers 32%; rapists 38%) or by a foreign object (sexual
murderers 11%; rapists 6%). Penetration or touching of the victim’s anus was rare for both
groups.
4.4.1.8 Mental Health
Table 4.1 displays the sexual murderers and rapists scores on the MCMI. The range of scores
obtained by each group is included as well as the mean scores for each group, and the
percentage of offenders whose scores were suggestive of clinical or personality disorder
(based on the cut-off of 70). Approximately 50% of the rapists and 60% of the sexual
murderers scored above the clinical cut off on the Anxiety scale. Antisocial Personality
Disorder was suggested for one-third of both groups, with alcohol dependence indicated in
76
one-third of the rapists and one-quarter of the sexual murderer samples. Approximately one-
third of both the sexual murderers and the rapists also indicated scores above clinical
significance for the Depressive scale. While none of the differences in scores between the
groups were significant, overall more of the rapists had scores suggestive of psychopathology
than the sexual murderers.
77
Table 4.1
Range of Base Rate Scores, and Percentage of Rapists and Sexual Murderers with Raw Scores Greater than 74 and Greater than 84 on the
Personality and Clinical Syndrome Scales of the Millon Clinical Multiaxial Inventory-III (MCMI-III)
Murderers
Total Sample a
Rapists Sexual b
Range
Mean Mean
(Base Rate 75-
Score Score
MCMI-III Scale Score) 84 >84 75-84 >84 Fc p ƞ2
Schizoid 0 - 88 43.42 8.4 2.8 42.64 3.6 0.0 0.01 0.92 0.00
Avoidant 0 - 95 37.06 14.0 5.6 38.46 17.9 0.0 0.04 0.85 0.00
Depressive 0 - 111 56.22 16.8 11.2 53.14 28.7 7.2 0.17 0.68 0.00
Dependent 0 - 90 40.72 8.4 2.8 41.61 10.8 7.2 0.02 0.90 0.00
Histrionic 0 - 81 50.53 5.6 0.0 50.25 7.2 0.0 0.00 0.95 0.00
Narcissistic 10 - 96 54.36 5.6 5.6 52.14 3.6 0.0 0.36 0.55 0.01
Antisocial 0 - 96 56.24 25.2 5.6 59.93 21.5 10.7 0.38 0.54 0.01
Sadistic 0 - 104 42.19 2.8 5.6 42.57 3.6 0.0 0.00 0.95 0.00
Compulsive 0 - 99 51.67 11.2 2.8 50.82 10.8 0.0 0.03 0.86 0.00
Negativistic(passive-
0 - 96 33.75
aggressive) 8.4 11.2 33.14 10.8 7.2 0.01 0.94 0.00
Masochistic (self-defeating) 0 - 93 39.89 14.0 2.8 41.75 14.4 7.1 0.06 0.81 0.00
Schizotypal 0 - 86 26.89 8.4 2.8 35.39 0.0 0.0 1.21 0.28 0.02
Borderline 0 - 95 45.28 11.2 8.4 43.5 3.6 3.6 0.08 0.78 0.00
Paranoid 0 - 92 39.36 2.8 8.4 28.64 0.0 3.6 1.99 0.16 0.03
Anxiety 0 - 102 52.92 25.2 19.6 58.61 42.9 17.9 0.44 0.51 0.01
Somatoform 0 - 71 33.31 0.0 0.0 20.11 0.0 0.0 3.35 0.07 0.05
Bipolar: manic 0 - 99 47.58 5.6 0.0 42.21 0.0 3.6 0.83 0.37 0.01
Dysthymia 0 - 104 41.19 11.2 2.8 45.71 7.2 0.0 0.37 0.54 0.01
Alcohol dependence 0 - 105 57.19 22.4 8.4 57.36 10.8 14.3 0.00 0.98 0.00
78
Drug Dependence 4 - 92 57.08 16.8 5.6 59.46 17.9 3.6 0.22 0.64 0.00
Posttraumatic stress disorder 0 - 78 44.19 5.6 0.0 45.86 7.1 0.0 0.06 0.80 0.00
Thought disorder 0 - 81 29.28 2.8 0.0 32.39 0.0 0.0 0.21 0.65 0.00
Major depression 0 - 103 39.36 2.8 5.6 29.18 0.0 0.0 2.06 0.16 0.03
Delusional disorder 0 - 104 29.31 0.0 2.8 31.07 0.0 0.0 0.06 0.81 0.00
a. N=36
b. N=28 Overall F df df p ƞ2
c. df=1 1.15 1 62 0.34 0.42
79
4.4.1.9 Summary of Comparisons
The rapists and sexual murderers were found to differ significantly on only the minority of
variables investigated. In terms of how the offence was committed, sexual murderers more
often caused serious injury to their victims. In contrast, the rapists were more likely to use
coercion. Contact with or penetration of the victim’s vagina with the offender’s penis was
more common in the rapist sample. With regard to the victims, the sexual murderers targeted
older women compared to the rapists. Overall, the rapists and sexual murderers differed
significantly on only a very small proportion (6 out of 78 comparisons; 8%) of the variables
investigated (see Tables 1, 2 and 3 for overview of comparisons), lending support to
combining the two groups to form a continuum of sexual violence.
4.4.2 Pathways to offending
Having established that the rapists and sexual murderers were sufficiently similar to
be further studied as one group of sexual aggressors, we investigated whether the offenders in
our sample would form similar clusters to those reported in previous studies (e.g., Proulx &
Beauregard, 2009 a, b; Beech, et al., 2005; Beech, et al., 2006). All of the variables
mentioned in the previous section of this chapter were entered together into one
agglomerative hierarchical cluster analysis. This method of clustering was chosen as it starts
with each case as single clusters and then repeatedly merges the two closest clusters until a
single, all encompassing cluster is left (Milligan & Cooper, 1987). Ward’s method was used
as it assumes that a cluster is represented by its centroid and it attempts to minimize the sum
of the squared distances of points from their cluster centroids (Tan, Steinback, & Kumar,
2005) and has been found to be robust and consistently better for recovering clusters from
bivariate data as well as multivariate data (see Milligan & Cooper, 1987; for a review of
validation studies). This analysis resulted in three distinct clusters. Table 4.2 shows the
percentage occurrences for all the personality profile, pre-crime, sexual lifestyle, and modus
operandi factors across the resulting cluster groupings.
80
Table 4.2
Personality, Pre-crime, Sexual lifestyle and Modus Operandi across Offending Pathway
Angry Sadistic Compensatory
(n=35)* (n=28)** (n=39)***
Personality a
Schizoid 45.1 39.1 44.2
Avoidant 37.6 33.7 40.9
Dependent 39.1 35.6 47.4
Histrionic 50.6 55.3 46.4
Narcissistic 53.1 60.2 48.4
Antisocial 51.8 61.4 61.1
Obsessive-Compulsive 55.2 54.6 44.8
Passive-Aggressive 28.2 31.0 40.7
Schizotypal 25.7 33.9 32.9
Borderline 39.5 46.1 48.3
Paranoid 27.7 37.3 39.7
Angry Sadistic Compensatory

(n=35) (n=28) (n=39)
Pre-Crime
Any previous 71.4 67.9 87.2
Redundancy 5.7 0.0 12.8
Relationship breakup 48.6 57.1 38.5
Sexual problems 0.0 3.6 0.0
Idleness 31.4 64.3 61.5
Familial problems 20.0 28.6 30.8
Sexual Lifestyle
Pornographic material 0.0 3.6 5.1
Prostitutes 11.4 10.7 5.1
Compulsive masturbation 2.9 35.7 0.0
sexual dissatisfaction 31.4 92.9 20.5
Number of sexual partners 45.7 32.1 46.2
Deviant fantasies 0.0 50.0 23.1
Modus Operandi
Deviant sexual fantasy
(48hrs) 8.6 50.0 89.7
Pre-crime anger 45.7 78.6 66.7
Explicit planning 14.3 60.7 59.0
Explicit victim selection 0.0 21.4 28.2
Victim prostitute 8.6 7.1 2.6
Restraints 25.7 82.1 23.1
81
Kidnapped/forced 5.7 14.3 17.9
Humiliation 11.4 10.7 7.7
Mutilation 2.9 0.0 0.0
Death of victim 57.1 53.6 7.7
4.4.2.1 Angry cluster
The first cluster contained 34% of the sexual aggressors. A large number of these individuals
had a prior history of criminal offences, a degree of sexual dissatisfaction, a large number of
sexual partners, and used prostitutes. They experienced a moderate degree of pre-crime
anger, and often restrained their victims during the offence, some of which were prostitutes.
They humiliated, and in some instances, mutilated their victims, and they had the highest
level of victim death of the three clusters. Obsessive-Compulsive personality disorder was the
most common personality disorder reported for this clustering of sexual aggressors, followed
by Narcissistic and Antisocial Personality Disorders.
One offender in this cluster commented that he left home in the evening to go
drinking, “looking for a fight”. Having got thrown out of one pub, he hitched a ride home
during which he sexually assaulted the victim. When she retaliated and hit him, he reported
losing his temper, being physically and sexually violent towards her until he killed her. He
then disposed of the body, leaving it bound and went home. Another man in this group
reported that he started to resent and hate his victim and that by committing the offence he
could take his feelings of anger and frustration out on his victim. These accounts illustrate
the anger and frustration that often preceded the offences committed by the sexual aggressors
within this pathway.
4.4.2.2 Sadistic cluster
The second cluster represents 28% of the sexual aggressors in the sample. They reported
their lives prior to the offence to be unstructured and inactive, and often they had experienced
some sort of relationship break-up leading up to the offence. What is particularly striking
about this cluster are their levels of compulsive masturbation and sexual dissatisfaction. Half
of these offenders engaged in deviant sexual fantasies, and just prior to their offence, they
reported that these fantasies became more elaborate. Almost 80% of the offenders in this
cluster reported experiencing anger during the 48 hours preceding the offence. These
offenders explicitly planned their offence, both restraining and humiliating their victims. Just
82
over half of the group’s violence toward the victim resulted in her death. Antisocial and
Narcissistic Personality Disorders were the dominant personality traits, reaching relatively
highly scores (60+). Histrionic traits were also evident in this clustering of sexual aggressors.
Self-centredness, ideas of grandiose self-importance and envy of others was expressed as
a motivator by one offender in this cluster. He reported being jealous and envious of what
other men who were going out with women had. He also articulated sexual dissatisfaction
prior to the offence “I hadn’t had sex for a while, for quite a long time”. Another offender
describes very candidly being sexually aroused and having deviant sexual fantasies
concentrated around thoughts of sex with young girls and thoughts of raping a stranger
female, then killing her, and then raping her after death. These fantasies he regularly
masturbated to and which formed the ‘plan’ for his victim.
4.4.2.3 Sexually Compensatory cluster
The third cluster represented the final 38% of the sample. The majority of offenders in this
group had previous convictions (87%), the largest percentage out of the three clusters. They
reported experiencing redundancy and familial problems leading up to their offence. They
explicitly planned their offences, including victim selection, and often forced or kidnapped
their victim to go with them. The majority of these offenders engaged in elaborate deviant
sexual fantasies within days of their offence, although deviant sexual fantasies were not a
consistent part of their general lifestyle. While they experienced a moderate degree of pre-
crime anger, this group had the lowest level of victim death. Approximately half of this group
of sexual aggressors had a moderate score for Borderline, Narcissistic and Antisocial
Personality Disorders.
One offender explains how he gained entry into his victim’s home to commit the
planned sexual assault: “and knocked on the door, and the lady came to the door, and I
tricked my way in, said that I was a (regional) man and that there was something going on in
the street and I tried to trick my way in.” During an assault against a known victim, one
offender commented to her: "how can this be rape if it's giving you pleasure?" and assaulted
her on a second occasion in which she had to fake orgasm before he would stop.
83
4.5 Discussion
4.5.1 Comparison of rapists and sexual murderers
As has been reported in previous studies (e.g., Grubin, 1994; Oliver et al., 2007), a greater
proportion of rapists were of an ethnic minority background, while the sexual murderers were
more often White. This observation is similar to that found in Greenfield’s (1997) sexual
murderer sample, where sexual murder was more often committed by a White offender. In
the current study, the rapists were more likely to be in a long-term sexual relationship at the
time of the offence, and fewer reported experiencing difficulty making and retaining
friendships compared to the sexual murderers. This supports previous findings that sexual
murderers tend to be more socially isolated in comparison to rapists. It could be argued that
the murderers see sexual offending as more appealing because it does not place demands on
their self-confidence or their deficient social skills (Beauregard, Stone, Proulx, & Michaud,
2008; Pardue & Arrigo, 2008) and accommodates their feelings of inadequacy (Groth &
Birnbaum, 1979).
Consistent with Beech et al. (2005), Langevin (2003), and Oliver et al. (2007), the
sexual murderers committed their first offence at a younger age, although this difference did
not quite reach significance. This highlights the importance and the need to detect early
warning signs in adolescent sexual offenders (Langevin, 2003) before they have the
opportunity to escalate further down the continuum of sexual violence. The victims of the
sexual murderers tended to be older than rapists. This might be because the sexual murderers
purposely target older victims because they are perceived as weak, vulnerable, and more
available, similar to sexual murderers of children (Beauregard, et al., 2008). Another
possibility is that the age of the sexual murderers’ victims makes them more likely to
succumb to their injuries.
Despite the fact that both groups showed an elevated occurrence of avoidant,
depressive, and antisocial personality traits, along with anxiety, and alcohol/drug dependence,
overall they did not significantly differ on their personality correlates. These findings are
similar to those of Oliver et al. (2007), Proulx et al. (2002), and Proulx et al. (2007), which
show that “rapists and sexual murderers have extremely similar personality styles and similar
prevalence of personality disorder” (Oliver et al., 2007, p. 310). This highlights the more
disturbed personality make up of these offenders in general.
Similar to Salfati and Taylor (2006), the rapists in this UK sample more often brought
a weapon with them to the offence, used instrumental violence, and were more likely to
84
penetrate their victim’s vagina with their penis rather than a foreign object. In contrast, the
sexual murderers more often used weapons taken from the scene of the offence, used a level
of violence that was beyond that needed to commit the assault, and although not significant,
more often engaged in foreign object insertion. This supports the argument that rape and
sexual murder occupy separate ends of a continuum of sexual violence. The difference
between a sexual assault and a sexual murder seems more to do with the situation in which
the offence takes place than the developmental, psychological, or criminological factors of
the offender (Beech, et al., 2005).
4.5.2 Pathways to Offending
The works of Proulx and Beauregard, as well as other research (e.g., Beech, et al., 2005;
Beech, et al., 2006), suggests three or four pathways to sexual offending, namely, the
Sadistic, Angry, Sexually Opportunistic, and in some instances, the Compensatory. It is clear
that the current clusters can be seen to represent these typical typologies. It can be argued that
Cluster 1 and 2 share similarities with the Angry and Sadistic typologies. While, Cluster 3
has similarities found to be related to a Compensatory pathway.
The first group of Angry sexual aggressors (cluster 1), similar to Proulx and Beauregard
(2009a), experienced anger just prior to the crime, and humiliated their victims, who were
sometimes prostitutes. The act of sex for these offenders is a way to further punish and
humiliate their victims and is not just for sexual gratification (Pardue & Arrigo, 2008). The
resulting death of their victims could be linked to the offenders’ aim of physically hurting
their victim and the use of expressive violence throughout the offence (Groth & Hobson,
1997; Pardue & Arrigo, 2008; Proulx & Beauregard, 2009a), as their aggression often spans a
wide range from verbal abuse to murder (Knight & Prentky, 1987). Again, similar to existing
descriptions (e.g., Beauregard, Proulx, & St-Yves, 2007), they rarely engaged in planning
their crimes and did not plan whom they were going to offend against, as their attacks are
often precipitated by or induced by life circumstances (Pardue & Arrigo, 2008). They also
reported lower levels engagement in sexual fantasies just prior to their offence. While none of
the personality disorders reached significant levels, this group of offenders had moderate
scores on Obsessive-Compulsive and Narcissistic Personality Disorder scales, indicating a
fairly anxious individual who may be interpersonally exploitative of others (DSM-IV-TR
American Psychiatric Association, 2000).
85
The Sadistic sexual aggressors in cluster 2 had moderate levels of relationship
dissatisfaction, and relational problems, possibly resulting in their high level of pre-crime
anger, and reported deviant sexual fantasies during the hours prior to their offence. Such
fantasising may have aided in establishing an offence-script (Pithers, et al., 1988; Ward &
Hudson, 2000) as the crimes of these offenders had been planned. Further, the excitement for
the sadistic aggressor comes from the infliction of pain upon and the humiliation of their
victim, which is meticulously planned beforehand, often in the form of elaborate sexual
fantasies (Deu & Edelmann, 1997; Langton & Marshall, 2001). Consequently, this type of
offender will stalk, abduct, and abuse their victim. Often they will murder their victims as a
precaution against detection (Groth & Birnbaum, 1979). Thus, the sexual attack, and
sometimes the subsequent murder, are planned and intended to inflict pain and cause terror
for the victim. There is often prolonged contact with the victim, which can last hours or even
days with the victim also being chosen based on the offender’s fantasies. The offender’s
appetite for gratification and the killing process is fulfilled through the torture and mutilation
of their victim (Keppel & Walter, 1999).
Many of these factors are apparent in the Sadistic pathway found in the current sample of
sexual aggressors. For instance, they used restraints, force, and humiliated their victims, often
resulting in the death of their victim, which is in line with Proulx and Beauregard (2009a,b)
and other studies (i.e., Beauregard & Proulx, 2002; Beech, Fisher & Ward, 2005; Brittain,
1970; Dietz, Hazelwood & Warren, 1990; Gratzer & Bradford, 1995; Meloy, 2000; Warren et
al., 1996). Personality scores of this cluster group were consistent with a highly narcissistic
and antisocial individual, who had little concern for the feelings or needs of other people,
exploited others to achieve their own goals (e.g., sexual satisfaction through victim
suffering), and who showed a lack of remorse or empathy when doing so (DSM-IV-TR
American Psychiatric Association, 2000). These were not consistent with Proulx and
Beauregard’s (2009a, b) sadistic pathway whose main personality disorders were avoidant,
schizoid, and dependent. This could be a result of different criteria or methods for
determining group membership.
The final group of sexual aggressors, the Compensatory (cluster 3), explicitly planned
their crimes and whom they were going to offend against. They rarely humiliated their
victim, although they did kidnap or forcibly move the victim to where the offence took place.
In order to affirm their masculinity they exhibit strength and control over their victims, this
also serves as a means for the offender to accommodate for their feelings of inadequacy
(Groth & Birnbaum, 1979; Knight, 1999), and to compensate for their perceived lack of
86
positive self-image (Pardue & Arrigo, 2008). They had the lowest rate of victim death,
indicating the use of non-lethal aggression during their sexual attack (Groth & Birnbaum,
1979). They also fantasised about their offence in the hours preceding it, which perhaps
involved more manipulative rather than aggressive or coercive sexual behaviours,
considering the possibility that they used non-lethal force. Often the assault is an expression
of these fantasies (Prentky, Burgess, & Carter, 1986). A large majority of them had previous
criminal convictions, were redundant, and reported idleness, indicating a generalised conflict
with society. They had a large number of sexual partners, which could be related to their need
for intimacy and continual reassurance of their masculinity (Pardue & Arrigo, 2008).
Borderline, Antisocial, and Narcissistic Personality Disorders, while not clinically
significant were present in this group of offenders to a moderate degree. Indicating the
individuals within this group have unstable interpersonal relationships (apparent in the
current cluster), a de-valued self-image, with unstable emotions and feelings about
themselves and others, and the tendency to violate the boundaries and rights of others (DSM-
IV-TR American Psychiatric Association, 2000). Overall, for this group, these findings do
not reflect the offending behaviour of either of Proulx and Beauregard’s groups of sexual
offenders. Although, these results do reflect the finding from studies on the power rapist
described by Groth and Birnbaum (1979), the power-reassurance described by Keppel and
Walter (1999), and the sexual non-sadist rapist of Knight and Prentky (1990), all of which
can be seen as different forms of the compensatory sexual offender.
4.6 Conclusions
Comparisons of rapists and sexual murderers in previous research and within the current
sample have all resulted in few differences being uncovered. These findings suggest that a
single model of escalating sexual aggression against adult females, often centering on the
level of coercion used to control the victim, would apply to both sets of offenders. In light of
these findings, rapists and sexual murderers were combined into a single group of sexual
aggressors against women and the relationship between different pre-crime, per-crime, and
post-crime factors and their pathways to offending investigated. Agglomerative hierarchical
cluster analysis of pre-crime, sexual lifestyle, modus operandi, and personality variables
resulted in a three-cluster solution. The three clusters of offenders, labelled ‘Sadistic’,
‘Angry’, and ‘Compensatory’, were comparable, but not completely consistent, with those
87
found previously in studies by Proulx and Beauregard (2002; 2009a, b) and others (Beech, et
al., 2005; Dietz, et al., 1990; Gratzer & Bradford, 1995; Warren et al., 1996).
The differences between the findings reported with this UK sample and those reported
in existing studies (Gratzer & Bradford, 1995; Grubin, 1994; Langevin, 2003; Langevin et al.,
1988; Milsom et al., 2003; Oliver et al., 2007; Proulx et al., 2007) could be the result of
different methods used to collect the data (i.e., different interview methods/questions; the file
information collected). In addition, the present data was collected from a sample of offenders
engaged in a treatment programme. The majority of the data was reliant on self-reported
information and it is possible that offences were misremembered and details omitted, or the
interviewer may have been given false or misleading information. However, a strength of the
current study was that all of the victims of the sexual aggressors were post-pubescent
females.
Interpretation of the clusters was complicated by the limitations of the data and the
variables available for the current sample. Some variables from the current data set had to be
merged to more closely represent those variables present in Proulx and Beauregard’s (2009a)
studies. For example, it was not possible to include pre-crime factors concerning the
offender’s self-esteem or conflicts with women in the analysis because this information was
not available, yet these were three variables that contributed heavily to Proulx and
Beauregard’s interpretation of their clusters.
Regardless, of these limitations, there was still some consistent findings between this
and previous studies which lends support to the merging of rapists and sexual murderers into
a single model of sexual aggressors against women, as well as their pathways to offending.
The usefulness of identifying supported pathways of offending for a combined sample
of sexual aggressors against women can be seen in their use with regards to criminal
investigative advice and analysis. The sexual offence is a process, or pathway, that develops
over time, with many different influencing elements affecting each other at different stages of
the offence process. Therefore, in order to understand the offence, and thereby the offender,
the complete process or pathway of the offence needs to be considered. Understanding and
investigating sexual aggression using pathways allows for the inclusion of both the ‘person’
and ‘situation’ component of the offence, as well as, a detailed description of what happened
with regards to behaviour, feelings, motivations, and situational characteristics, before,
during, and after the offence (Crabbe, Decoene, & Vertommen, 2008). These can then be
used to help focus police investigations by providing relevant leads and strategies, and
88
eliciting, identifying and prioritising potential suspects (Mokros & Alison, 2002; Warren, et
al., 1999).
4.6.1 Suggestions for future research
In future research, it would be beneficial to include more dynamic and changeable factors
(i.e., beliefs, attitudes, cognitions, motivations) within the categories (i.e., pre-crime,
general/sexual lifestyle) tested in the present study, especially as these factors are more
amenable and readily targeted in treatment, allowing for the development of more accurate
risk assessment (Fisher & Beech, 2007). Another area needing further study is the influence
of situational factors on the perpetration of aggressive sexual offences against women as
studies of escalation have found situational factors such as alcohol consumption, weapon use
and victim resistance to be indicative of an increase in the severity of the sexual assault
(Beauregard, Lussier & Proulx, 2009; Ch ѐnѐ & Cusson,
Pullman, 2007;
1998). The
offender’s behaviour should be considered in terms of both his underlying psychological
processes and the situation and context in which the offence actually occurred (Crabbe, et al.,
2008; Douglas, et al., 1992; Goodwill & Alison, 2007) as this interaction may influence the
final result of the offence (i.e., death of victim).
89
Chapter 5, 6, and 7: The Potential Mediators between Crime Scene and Offender
Characteristics
Chapters 5-7 build on the factors discussed in Chapter 2 with regards to intermediary factors
within the A to C equation of Offender Profiling. Chapter 5 looks at five contextual variables,
the use of drugs just prior to the offence, the use of alcohol just prior to the offence, location
of initial contact with victim, location of actual offence, and the level of victim resistance,
and the degree and nature of their influence when inferring offender characteristics from
crime scene characteristics. Chapter 6 examines the occurrence and influence of implicit
theories and the information they provide about the potential views the offender held about
themselves, the people around them, and their world at the time of their offence. This chapter
also provides the first part of the psychological make-up of the offenders discussed in the
current thesis. The second part exploring the psychological make-up of the offender is
Chapter 7, in the form of the potential motivations of the offender found in previous research:
anger, sadistic, sexually opportunistic, and sexually compensatory. These provide a further
exploration of the psychological meaning of the offender’s crime and their motivations for
sexually offending. The three chapters combined look at the potential relationships between
the contextual and cognitive-affective aspects of the offender and their sexual offending
behaviour.
90
Chapter 5: Context as a Mediator of the Interaction between Crime Scene and Offender
Characteristics
5. Introduction
Offender profiling infers the relationship between offender characteristics from crime scene
aspects. There has been a great deal of research done over the last three decades, some testing
the accuracy of this assumption with mixed results, but the majority of what has been
published with regards to offender profiling have been discussion pieces, lacking a clear
theoretical framework leaving the underlying processes of offender profiling still not well
understood (see Dowden, Bennell, & Bloomfield, 2007 for overview). Often, the relationship
between crime scene and offender characteristics are tested as item-to-item comparisons,
although there is more current research which indicates that this relationship may actually be
influenced by the circumstances surrounding the offence. One example is Goodwill and
Alison (2007) who found the relationship between offender characteristics (i.e. age) and
crime scene information (i.e., victim age) to be moderated by other offence characteristic(s)
(i.e., planning; level of aggression). Models from personality psychology, outlined in Chapter
2, would also suggest that the relationship between crime scene and offender characteristics is
unlikely to be straight forward as originally thought (or hoped). This chapter will seek to
explore possible contextual/situational factors that may influence the relationship between
offender characteristic and offence characteristics during a sexual offence.
It is suggested by the Cognitive-affective personality system (CAPS) theory (Mischel
& Shoda, 1995) that situation and context interact with psychological and personality aspects
of a person to produce behaviour (Mischel, 1999). Under this theory the focus is to identify
situational features that are relevant for the person whose behaviour we are interested in (Van
Mechelen, & De Raad, 1999). The variability of behaviour seen across different situations, to
some extent, may be the expression of the interplay between stable personality structure and
the unpredictability of the situation (Mischel, Shoda, & Mendoza-Denton, 2002). Relating
this to sexual offending, it can be postulated that different situational factors, such as the use
of drugs or alcohol or victim resistance, will interact with the personality traits of the
individual offender to produce certain offence behaviours (e.g., excessive aggression; use of a
weapon).The offender may exhibit a number of variant behaviours during their crime,
depending on the situation and the psychological features of the situation, but these will be
distinctive and stable patterns of situation-behaviour relationships (Mischel et al., 2002;
91
Salfati & Bateman, 2005). Therefore, cross-situational variability is the exhibition of the
flexible personality of the constructs and its stable underlying organisation characterized by
if...then...relationships (Mischel et al., 2002). When the situationally determined ifs change so
do the personality determined thens, although the if...then...relationships remain the same
when the pairings remain the same (Mischel et al., 2002). The goal of the theory is to be able
to make specific predictions about how people are likely to think, feel, and behave in and
across certain kinds of situations (Mischel, 1968).
For the purposes of this study, and the thesis, the contextual factors chosen to be
explored are 1) the location of the initial contact with the victim; 2) the location of the
offence; 3) the use of drugs or alcohol just prior to the offence; and 4) the level of victim
resistance. All of these have empirical support for playing significant roles in the process of a
sexual offence, although the support for their influence on the offender’s personal
characteristics is less conclusive and abundant.
5.1 Location of Initial Contact and Offence
According to Canter (2004), one key area for the inferential process of offender
characteristics from crime scene aspects is the environmental psychology of the crime.
Within the subtext of the geometric and environmental theory, crime can be seen as an event,
which occurs in a specific location and in a specific situation (Brantingham & Brantingham,
1981; Felson, 1987). There are a variety of factors, such the suitability and risk level of being
caught, that influence the attractiveness and accessibility of an area or location; the
opportunities to commit crimes; target selection; and the potential offence behaviours of
offenders. Target selection, or the location of initial contact with the victim, may be
influenced by factors more related to the characteristics of the area itself (e.g., isolation,
number of potential victims/opportunities), while the actual crime location may be influenced
more by the activity level and awareness of different geographical areas by the offender
(Canter & Larkin, 1993; Frank, Andresen, & Brantingham, 2011; Frank, Dabbaghian, et al.,
2011). The offender’s criminal as well as legitimate and non-criminal movement around their
environment (e.g., employment, school, social interaction) will influence both of these
(Blackburn, 1993; Brantingham & Brantingham, 1981).
Furthermore, the individual differences between offenders (e.g., SES, age, sex)
(Brantingham & Brantingham, 1981; Canter & Alison, 2000; Heth, Cornell, & Flood, 2002;
Matthews, 1995) will play a part in the location of both initial contact and actual crime
92
location. Employment, socio-economic status (SES), and age, have been shown to influence
the distances and locations travelled to by individuals often depending on their mode of
transportation, the availability of amenities, and the location of resources (Mercado & Paez,
2009; Morency, Paez, Roorda, Mercado, & Farber, 2011). These factors (low SES, unstable
employment, and younger age) have also been linked with persistent general offending
(Gendreau, Little, & Goggin, 1996), and have been found in sexual offenders as potential risk
factors (Polaschek, Ward, & Hudson, 1997; Simon, Sales, Kaszniak, & Kahn, 1992). This
demonstrates, at the least, the potential links between an offender’s demographics, their
movement around particular and potential crime locations.
The psychological importance of the locations has also been found to have an
influence on the location of sexual offences (Meaney, 2004; Mischel et al., 2002; Salfati &
Bateman, 2005). For example, Canter and Larkin (1993), Kocsis and Irwin (1997), and
Meaney (2004) found that serial sexual offenders tended to adopt a marauder style of
offending, branching out to offender within a certain proximity to their home base, as
opposed to travelling far distances to commit their crimes. This suggests a strong link
between the location of the offender’s home and their spatial behaviour (Canter & Larkin,
1993), as well as their offence site, and victim selection (Meaney, 2004).
Sexual offences can be motivated by interpersonal violence, with many opportunities
for such to happen within any given location (Kocsis & Irwin, 1997), however, offenders
often look for targets within a more restricted space. It is the interaction of the location of
potential targets and the offender’s awareness of that location that results in particular crime
behaviour (Brantingham & Brantingham, 1984). Offenders learn from previous experiences
and social interactions where and who would be ‘good’ targets for their offences
(Brantingham & Brantingham, 1981), and frequently these locations will have particular
significance or symbolic salience for the offender (Coucelis, Golledge, Gale, & Tobler,
1987). Often, these locations are those the offender frequents during their daily lives and
routine, and they dictate the location and vulnerability of personal and property targets
(Blackburn, 1993). Once a suitable hunting ground or offence location has been identified,
and a suitable target chosen, there is still the necessity that any persons who would seek to
stop or interfere with the offender is absent from the location before the offence to occur
(Cohen & Felson, 1979; Clarke & Felson, 1993).
It has also been found that offence outcomes can differ depending on the location of
the offence. Plye (1974) found that rapes tended to occur outdoors, in areas of urban renewal,
temporary lodging, and construction. Canter and Larkin (1993) found that crime locations
93
chosen by a sample of serial sexual offenders adhered to a precise, circular region, with the
vast majority of their sample living within this circular area. They suggest that there is a
psychological basis for the offender’s choice of crime location. Once the location of the
sexual offence has been determined, another potential influencing factor of the offence
outcome is the behaviour of and the interaction between the victim and the offender.
5.2 Victim Resistance
Another situational aspect proposed to influence how an offender commits their sexual
offence is that of the victim-offender interaction, more specifically, victim resistance.
Research suggests that outcomes of aggressive and violence interactions are at least partly a
function of the events that occur during the incident (Felson & Steadman, 1983). It has been
hypothesised that verbal and physical resistance by the victim may be correlated with verbal
and physical attacks by the offender and that the more aggressive the victim the more
aggressive the response from the offender. This interaction may be influenced by the
presence of a weapon (Felson & Steadman, 1983), which may work in one of two ways: 1) to
subdue the victim into compliance or 2) increase the amount of victim resistance. It has been
shown that the more the victim resists there is a greater possibility of resulting severe injury
to themselves (Block & Skogan, 1986; Ullman, 1998), and of greater sexual victimisation
severity (Ullman, Karabatos, & Koss, 1999a), with the possibility of being killed rising as a
result (Felson & Steadman, 1983).
Conversely, it has also been found that victims who respond with forceful physical or
verbal resistance reduce the likelihood of a completed sexual assault (Block & Skogan, 1986;
Ullman, 1998; Ullman & Siegel, 1993), without increasing their risk of physical injury
(Kleck & Sayles, 1990; Ullman & Knight, 1992), especially in the instance of offenders using
verbal pressure as opposed to physical force (Ullman & Siegel, 1993). According to Canter
(1996):
It is the variety of actions that happen in sexual attacks that indicate the different
modes of relationship that offenders have with their victims. Any empirical model of
offence behaviour must therefore encapsulate and explicate these variations in mode
of interaction with the victim. (p. 192)
Therefore, in order to fully understand what has occurred during the course of a sexual
offence the interactions between the offender and their victim need to be examined and
94
understood, with victim resistance being one aspect of this interaction. An example of the
influence of victim resistance on offence outcome linked with offender perception was the
work of Fritzon and Ridgway (2001). They found that how the offender perceived the victim
would affect how the offender reacted if the victim resisted. The offender places their victim
in certain roles, which are based on their previous interpersonal relationships, and their
interactions with others (Canter, 1989). Depending on which role the victim takes will vary
the level of power and control exhibited which will be reflected and represented in the
corresponding crime scene actions (Canter, 1989). The different roles are the victim as a
significant person; as a vehicle to achieve some external goal; or as a depersonalised object,
each having their own implication for the levels of violence displayed by the offender if the
victim resists. The most serious reaction to victim resistance is for victims who are seen as
depersonalised objects, as here the victim resistance works to escalate the violence. For those
victims who are seen as significant people, the use of resistance is more beneficial as it was
more likely that the offender would try to change their behaviour, to try to calm the victim
down rather than becoming more violent towards them. The victim role as a vehicle was
found to have less of an influential role between offender and offence outcome.
The perception of the offender themselves, by the victim also plays a part in whether the
victim resists the attack or complies (Luckenbill, 1981). The victim assess the capability of
the offender being able to inflict serious injury dependent on whether the offender possesses
lethal resources and is in a position to use them. The capability of the offender to inflict
serious injury may be in the form of a weapon, such as the brandishing of a knife or gun, or
can be dependent on the appearance of age, strength, and physical size. If they victim does
not believe the offender is able to physically injury them, they may chose to resist. The
willingness of the offender to use such force is also assessed by the victim, in two ways,
whether they will use the force only in the face of victim resistance, or if they will regardless
of victim resistance (Luckenbill, 1981). Again, if the victim believes that the offender will
use lethal force if they resist they may choose to comply, whereas, if they believe the
offender will use lethal force regardless, the victim may then choose to resist even in the face
of serious injury. Therefore, if the victim believes the offender is capable and willing to
inflict serious injury and/or death in the face of victim resistance, then often the victim will
not resist and complies with the offender’s demands. From this point, the offender than
responds according to how they perceive the victim and the potential for further resistance
from the victim (Luckenbill, 1981). Such research suggests that victim resistance may be a
mediator for some offenders.
95
5.3 Alcohol and Drug Use
Alongside, the location of the offence and the victim-offender interaction, the use of drugs or
alcohol prior to the offence has also been shown to be interconnected between the offender
and his offence behaviours. There is evidence for the association between acute alcohol use
with aggression (e.g., Chermack & Blow, 2002; Murphy, Winters, O’Farrell, Fals-Stewart, &
Murphy, 2005), as well as acute drug use with aggression and violence, albeit less
substantively (e.g., Chermack & Blow, 2002; Davies, 1996; Goldstein, 1985). Alcohol has
been directly associated with the risk of sexual assault and is present in a large proportion of
sexual assault incidents (Abbey, 1991; Pernanen, 1991; Rajaratnam, Redman, & Lenne,
2000). Alcohol just prior to an assault has been shown to be strongly related to violence and
aggression (Busch-Armendariz, DiNitto, Bell, & Bohman, 2010; Bushman & Cooper, 1990;
Collins & Schlenger, 1988; Ullman & Knight, 1993), with a direct association between pre-
assault alcohol consumption by the offender and more severe sexual victimisation to the
victim (Ullman et al., 1999a). Results looking at different categories of violent criminals
(e.g., rapists; murderers; assault) have found similar results (Myers, 1986; Roslund & Larson,
1979; Tinklenberg & Ochberg, 1981; Wolfgang & Strohm, 1956).
Offenders’ pre-assault drinking has been related to greater rape completion (Brecklin
& Ullman, 2002) and to increased victim injury (Ullman & Brecklin, 2000). Although, it has
also been found that more severe sexual victimisation in sexual attacks was more likely in
instances where the offender did not use alcohol (Martin & Bachman, 1998), or no link has
been found at all between offender drinking prior to the attack, physical aggression against
the victim or likelihood of rape completion (Ullman, Karabatsos, & Koss, 1999b). Alcohol is
clearly associated with the outcome of sexual assaults, although its exact role in physical
injury still needs to be further explored, but it can be concluded that drinking alcohol is one
of many situational factors associated with sexual assault (Ullman, 2003) and the interaction
between offender and victim.
Links have also been found between offender demographics and alcohol
consumption. Alcohol consumption and criminal behaviour have been shown to be
interrelated, with greater usage found among juveniles and young adult offenders (Chermack
& Blow, 2002; Collins, 1981), often leading to further criminal behaviours (Weiner et al.,
2005) – signifying the development of a criminal history. As well, there have been links
found between the use of alcohol, the level of injury inflicted, and relationship status of the
offender (Chermack et al., 2010).
96
While drug use may be considered less important in sexual offending than that of
alcohol use (Langevin & Lang, 1990), the use of different types of drugs have been found to
affect different systems in the body (i.e., the Central Nervous System) (Rajaratnam et al.,
2000). Drugs have been found to be influential in facilitating, triggering, and causing
individuals to have violent outbursts (Goldstein, 1985; Weiner, Sussman, Sun, & Dent,
2005), altered mood states, and to enact criminal behaviours at the time (Rajaratnam, et al.,
2000), as well as result in future criminal offending (Weiner et al., 2005). This effect may be
exaggerated when drugs are used in conjunction with alcohol (Chermack et al., 2010).
Therefore, while alcohol use may be more prevalent among sexual offenders than nonsexual
violent offenders (Abracen, Looman, & Anderson, 2000), the use of drugs is still an
important factor to consider in the offence process.
5.4 Aim of Chapter
The focus of this chapter therefore will be on the following contextual variables, which are
hypothesised to have an influence on sexual offending: 1) use of drugs and/or alcohol just
prior to crime; 2) location of initial contact with victim; 3) location of the offence; and 4) the
level of victim resistance. As mentioned previously, these variables have been selected as
they have often been proposed to influence sexual and general offending, and have been used
in the development of some offender profiles. The general aim of the chapter is to explore
which, if any, of the chosen contextual variables will influence the relationship of inferring
offender characteristics from offence characteristics, and if these relationships can be used to
predict or generate a pragmatic list of offender characteristics for the investigation of the
offence and generation of a suspect list based on the available crime scene information, and
possible influences of situational variation.
97
5.5 Method 31
5.5.1 Sample
The sample consisted of 102 sexual aggressors (64 rapists; 38 sexual murderers) who had
committed, and were convicted, of at least one sexual offence against an adult (16 years of
age and above) female victim, and who were taking part in the Core SOTP between 1998 and
2002. For some of the analyses there was not complete information for all of the offenders.
For the perpetrator age analyses information was available for 67 of the 94 offenders’, except
for the perpetrator age and level of injury analysis which had a sample of 45 as there were
missing variables for level of injury within the dataset. Relationship status, any previous
convictions, lives alone, and employed analysis with level of injury had a sample of 56 due to
missing variables. The rest of the analyses had a sample of 94.
See Table 5.1 for the coding of the offence characteristics (X), the coding of the offender
characteristics (Y), and for the coding of the mediating variables (M). Each of these were
operationalized by the degree they could be evident from the actual offence and crime scene
information, as well as, the information contained in the offence details and the functional
analysis questionnaire.
31
See Chapter 4 for a full overview of sample and data collection methods. See Chapter 3 for a full overview of the statistical analyses used
in the current chapter.
32
See Chapter 3 for more detailed description of coded variables.
98
Table 5.1
Coding of Offence Characteristics (X), Offender Characteristics (Y), and Potential Mediators
(M)
1=minor injuries
3=major injuries
4=unknown
2=rape
Victim Age Interval; 16 to 86 years of age
1=used
Sexual Penetration 0=none/missing
Offender Characteristics (Y) Coding
Perpetrator Age Interval; 14 to 57 years of age
Relationship Status (in a sexual relationship at 0=no
time of offence) 1=yes
Any previous Convictions 0=no
1=yes
Lives Alone 0=’other’
1=alone
Employed 0=’other’
1=employed/student
Potential Mediators (M) Coding
Use of Drugs During the Offence 33 0=none
1=yes
Use of Alcohol During the Offence 0=none
1=yes
Location of Initial Contact between Victim and 0=indoor
Offender 1=outdoor
999=unknown
Location of the Offence 0=indoor
1=outdoor
999=unknown
Victim Resistance 0=nothing/unknown
1=pleading/trying to talk offender out of offence
2=shouting help
3=verbally hostile towards offender
4=physically hostile towards offender
33
There was no information available on the actual amount of drugs or alcohol taken; therefore, no indication of the actual level of
intoxication.
99
5.6 Results
One hundred and twenty different mediations were run looking at the relationship between
the five offender characteristics, four proposed mediating variables, and six offence
characteristics. The figures for these analyses can be found in Appendix D, along with the
tables of the Indirect and Direct Effects. Those that were significant will be discussed in the
following sections.
5.6.1 Multicollinearity
Multicollinearity was checked for all variables included in the present study. None of the
correlations between variables were above 0.40, the variance of inflation (VIF) ranged from
1.101 to 2.354, and the tolerance statistics (T) were between 0.425-0.903. These measures
indicate a low likelihood of issues related to multicollinearity affecting the regression model
(Field, 2009).
5.6.2 Mediation (Paths a and b), Indirect (ab), and Total Effects
Overall, there was one partial mediation found between the offender having previous
convictions (Y), the victim’s age (X), and the location of the offence (M) (Figure 5.1). No
other partial or full mediation was found. This partial relationship tentatively indicates that
the older the victim the more likely the offence took place inside and that offender had
previous convictions. As this was only a partial relationship, there is a direct positive
relationship between victim age and previous convictions in that as victim age increased so
did the likelihood that the offender had previous convictions. The partial mediation was
supported by the indirect effects (see Table 5.2).
-0.007** 1.345*
M
X Y
0.071*
Regression Summary: -2LL=94.724, Model LL=16.577, McFadden=0.149, Cox Snell=0.150, Nagelkerke=0.226
*p<0.05; ** p<0.01
Figure 5.1. Predicting Any Previous Convictions (Y) from Victim Age (X) through the
Location of Offence (M).
100
Table 5.2
Indirect Effects of Victim Age on Any Previous Convictions through Contextual Variables (ab
paths)
Bootstrapping
B SE Percentile 95% BC 95% CI BCa 95% CI

CI
Lower Upper Lower Upper Lower Upper
Drugs 0.000 0.010 -0.030 0.014 -0.027 0.017 -0.017 0.043
Alcohol -0.000 0.002 -0.005 0.004 -0.006 0.003 -0.006 0.003
Location of Initial
-0.004 0.006 -0.013 0.001 -0.014 0.001 -0.014 0.001
Contact
Location of Offence -0.010 0.010 -0.023 -0.001 -0.023 -0.001 -0.023 -0.001
Victim Resistance 0.000 0.003 -0.005 0.008 -0.005 0.009 -0.005 0.010
Note: BC, bias corrected; BCa, bias corrected and accelerated; 5,000 bootstrap samples
Indirect relationships were found between perpetrator age, drug use and the use of a
weapon during the offence (Table 5.3), and perpetrator age, drug use and the enactment of the
sexual penetrative behaviours (Table 5.4). If the offender had used a weapon there was a
greater likelihood they were under the influence of drugs and potentially more likely to be
older at the time of the offence. Again, the enactment of more sexually deviant penetrative
behaviours during the offence indicated a greater likelihood of drug use prior to the offence,
and combined were indicative of a potentially older offender.
101
Table 5.3
Indirect Effects of a Weapon Being Used on Perpetrator Age through Contextual Variables
(ab paths)
Bootstrapping
Product of
B Coefficients Percentile 95% CI BC 95% CI BCa 95% CI
SE Z Lower Upper Lower Upper Lower Upper
Drugs 0.392 0.500 0.784 0.015 4.339 0.002 3.833 -0.031 2.914
Alcohol 0.142 0.402 0.353 -0.866 1.206 -0.506 1.686 -0.490 1.762
Location of Initial
Contact 0.062 0.402 0.154 -0.970 1.156 -0.821 1.304 -0.878 1.227
Location of
Offence -0.031 0.171 -0.184 -0.456 0.942 -0.486 0.927 -1.632 0.533
Victim Resistance 0.051 0.190 0.271 -0.700 1.093 -0.491 1.449 -0.569 1.288
Table 5.4
Indirect Effects of Sexual Penetration on Perpetrator Age through Contextual Variables (ab
paths)
Bootstrapping
Product of
Drugs 0.202 0.217 0.932 0.018 2.071 0.010 1.894 -0.012 1.461
Alcohol -0.058 0.251 -0.231 -0.614 0.607 -0.709 0.501 -0.753 0.470
Location of Initial
Contact -0.070 0.167 -0.422 -0.583 0.281 -0.737 0.191 -0.777 0.185
Location of
Offence -0.017 0.068 -0.245 -0.239 0.293 -0.161 0.393 -0.403 0.212
Victim Resistance -0.026 0.079 -0.324 -0.376 0.263 -0.599 0.143 -0.599 0.143
102
Significant total effect 34 relationships were found for perpetrator age, level of injuries and
offence outcome (Table 5.5); relationship status and offence outcome (Table 5.6); previous
convictions, level of injuries and victim age (Table 5.7); and between the offender living
alone and level of aggression (Table 5.8). The analysis indicates that the relationship between
perpetrator age and level of injury was negative, therefore, as the level of injury increased the
perpetrator’s age decreased. The outcomes of the offence (rape or murder) also had a
significant relationship with perpetrator age, with rape indicating an older offender and
murder a younger offender, supporting the findings in Chapter 4.
It was significantly more likely that the offender was in a sexual relationship at the
time of offence if the sexual offence was a rape, whereas, if the offence ended with the death
of the victim then it was significantly more likely that the offender was not in a sexual
relationship at the time of the offence. A positive relationship was found between the level of
injury and previous convictions, indicating that as the level of injuries sustained by the victim
increased in severity so did the likelihood that the offender had a previous conviction. There
was a positive relationship between victim age and the offender having a previous conviction
indicating that as the victim age increased so did the likelihood that the offender had previous
convictions.
Table 5.5
Total Effects of Offence Variables on Perpetrator Age (c path)
Product of
Coefficients
B SE t p
Level of Aggression -0.626 1.312 -0.477 0.635
Level of Injuries -2.064 1.005 -2.054 0.046
Offence Outcome 5.479 1.963 2.791 0.007
Victim Age -0.039 0.054 -0.731 0.467
Weapon Used -2.771 2.444 -1.134 0.261
Sexual Penetration 0.182 0.985 0.185 0.854
34
The total effect is the sum of the direct effect of offence characteristics (X) on offender characteristics (Y; c’) and the indirect effect of
offence characteristics on offender characteristics (X on Y through M; ab).
103
Table 5.6
Total Effects of Offence Variables on Relationship Status (c path)
Product of
Coefficients
B SE z p Wald
Level of Aggression 0.038 0.237 0.158 0.874 0.025
Level of Injuries -0.268 0.243 -1.105 0.269 1.219
Offence Outcome 0.912 0.431 2.116 0.034 4.476
Victim Age -0.005 0.011 -0.426 0.670 0.181
Weapon Used -0.411 0.520 -0.789 0.429 0.624
Sexual Penetration -0.214 0.213 -1.005 0.315 1.011
Table 5.7
Total Effects of Offence Variables on Any Previous Convictions (c path)
Product of
Coefficients
B SE z p Wald
Level of Injuries 0.548 0.260 2.107 0.035 4.441
Offence Outcome 0.014 0.482 0.028 0.977 0.001
Victim Age 0.059 0.024 2.491 0.013 6.207
Weapon Used 0.174 0.619 0.282 0.778 0.080
104
Table 5.8
Total Effects of Offence Variables on Lives Alone (c path)
Product of
Coefficients
B SE z p Wald
Level of Aggression -0.583 0.275 -2.123 0.034 4.508
Offence Outcome -0.145 0.511 -0.283 0.777 0.080
Victim Age -0.036 0.020 -1.836 0.066 3.369
Weapon Used 0.110 0.628 0.176 0.861 0.031
5.6.3 Direct Effects of Offence Characteristics (X) on Offender Characteristics (Y)(c’) 35
There were two groups of significant c’ paths for perpetrator age: 1) level of injury and 2)
offence outcome. The relationship between perpetrator age and level of injury was a negative
one, indicating that as the offender’s age increased the level of injury to the victim decreased.
The relationship between offender age and offence outcome was a positive one, indicating as
the offender’s age increased so did the likelihood that the offence was a rape, as opposed to a
sexual murder.
There was one significant c’ path between relationship status and offence outcome,
although these were just on the border of significance (p=0.05). Again, as with the total
effect, and the findings in Chapter 4, this indicates that a rapist was more likely to be in a
relationship at the time of their offence, then a sexual murderer.
There were two groups of significant c’ paths for the offender having any previous
convictions: 1) level of injury and 2) offence outcome. These findings indicate that as the
level of injury increased so did the likelihood that the offender had previous convictions, and
that the offence ended in a rape only.
There was one group of significant c’ paths between the offender living alone and the
level of aggression used by the offender. This indicates that as the level of aggression
increased from instrumental to more than what was necessary to control the victim and
35
Path c’ is the direct effect of X (offence characteristics) on Y (offender characteristics) independent of the effects of M (Implicit
Theories).
105
commit their offence, it was more likely that the offender did not live alone and was co-
habiting. There were no significant c’ paths for the offender living alone or being employed
at the time of their offence.
5.7 Discussion
The aim of the chapter was to test for mediation between offender characteristics, offence
characteristics, and contextual variables. As can be seen from the results, there was only one
partially mediated relationship between these three groups of variables – victim age, the
location of the offence, and the offender having previous convictions. In addition, indirect
relationships were found between perpetrator age, drug use and the use of a weapon during
the offence; and perpetrator age, drug use and the enactment of the sexual penetrative
behaviours.
The influencing role that location plays in offending and the geographical analysis of
crime is not a new concept, and was independently discovered by different individuals at
different times (e.g., Brantingham & Brantingham, 1981; Kind, 1987; Rengert & Wasilchick,
1985; Rossmo, 2000). The geographical distribution of criminals and crime location was one
aspect of the crime environment that was hypothesised to be instrumental in making reliable
inferences between a crime scene and offender characteristics. The location of the actual
offence was significantly related to previous convictions and the victim’s age, which is in line
with previous findings. A 2007 Home Office Report (Feist, Ashe, Lawrence, McPhee, &
Wilson) found that over two-thirds of the rapes took place indoors (e.g., victim’s home,
offender’s home), along with the fact that older victims tend to live alone and be more
housebound (Walker, 1985) potentially steering the offence to happen indoors, and the
findings that repeat offenders tend to commit their victim-related offences indoors (Denno,
1986). Warren et al. (1998) findings also partly support the current findings, as they found
that those rapists with more extensive criminal histories, who travelled farther to offend,
often burglarised victims in their own homes, and were more likely to be white, and older.
Literature shows that drugs and/or alcohol can alter a person’s consciousness to
varying degrees, and that this may affect their behaviour, and more specifically, whether or
not they will commit a crime (Rajaratnam, et al., 2000). The results show that the use of a
weapon during the offence was an indication that the offender had used drugs prior. This
contrasts with Brecklin and Ullman (2001) who found no relationship between offenders’
weapon use and pre-assault alcohol use, and Resignato (2000) who found no strong link
106
between drug consumption and an increase of murder or other violent crimes. However, it
supports other findings that a large majority of violent offenders do report using drugs during
the same period of their crime (Chaiken & Chaiken, 1982), and that drug users (e.g., cocaine
users) exhibit more aggression compared to non users (Beachy, Petersen, & Pearson, 1979).
While, the current findings did not support previous findings with regards to the positive
relationship between pre-assault drug and alcohol use and level of injury suffered by the
victim (Ouimet, Guay, & Proulx, 2000), or the level of aggression exhibited by the offender
(Beauregard, Lussier, & Proulx, 2005), it may be hypothesised that the very use of a weapon
during the offence yields the potential for increased probability of injury and constitutes a
higher level of aggression, then say just using verbal coercion . The use of drugs prior to the
offence was also an indication of an older offender, which is in contrast to the majority of
research, which indicates a gradual decrease in drug use with age, although alcohol remains
more prevalent with age (Fendrich, Mackesy-Amiti, Goldstein, Spunt, & Brownstein, 1995).
This opposing finding could be the result of the ‘lumping’ of all the age categories of the
current sample of offenders together, as Fendrich et al.’s study suggests that categorical
analysis of age is needed. By not separating the age categories, the findings are only a
representation of the average use over the entire sample, and the actual links between age and
alcohol and/or drug use cannot be made as any the significant differences between the
different age categories are masked.
Drug use and perpetrator age were also associated with the more sexually deviant
behaviours. It may be that under the influence of drugs the offender enacts more sexually
‘deviant’ behaviours (e.g., anal sex) than that they would do, or could do, in their consenting
sexual relationships. It has been found that drugs, and alcohol, both act as sexual disinhibitors
and increase the likelihood of more risky sexual behaviours (Parker, Harford, & Rosenstock,
1994; Tapert, Aarons, Sedlar, & Brown, 2001). Although, with regards to offender age, it has
been found that juvenile or younger offenders report wider varieties of sexual deviant
interests then adult sex offenders (Zolondel, Abel, Northey, & Jordan, 2001). However, it
was also found that sexual recidivists self-report more varied deviant sexual fantasies
(Kenny, Keogh, & Seidler, 2001), and sexual deviancy has been found to be a risk factor for
adult sexual recidivism (Hanson & Bussiere, 1998). If sexual recidivists are older offenders,
this could explain the link between an ageing offender and their acting out of their deviant
sexual fantasies.
Surprisingly, victim resistance was not causally linked with any of the offence
characteristics or offender characteristics. Considering the support both for and against victim
107
resistance influencing the outcome of a sexual assault, a lack of significant finding within the
current data set is both expected and yet still perplexing. This could be related to the majority
the of victims in the sample using less physical means of resistance against their offender, as
outlined in Chapter 4. Trying to dissuade the rapist from committing the offence; crying
and/or pleading with the offender to stop, are behaviours requiring less physical reaction by
the offender.
The combination of the direct and total effects point to an interplay between the three
groups of variables even if there were not any mediated relationships. Specifically, the level
of victim injury was able to provide information about both the perpetrator’s age and whether
they had any previous convictions. For both of these relationships the inclusion of the current
potential mediating factors decreased the magnitude of this predictive relationship. Although,
it was still the case that greater victim injury meant a higher likelihood that the offender was
younger, and that the offender had previous convictions for various crimes. This is similar to
findings of Baxter, Marshall, Barbaree, Davidson, and Malcolm (1984), who found that
rapists who were younger at their index offence (compared to paedophiles or hebephiles)
were non-sexually and sexually more criminal, and frequently used violence and/or weapons
in their offences, increasing the potential to cause more serious victim injury.
The outcome of rape or murder was also informative about perpetrator age and their
relationship status – indicating that rapists tended to be older than sexual murderers, and were
more likely to be in a sexual relationship at the time of their offence. The inclusion of the
potential mediators increased the strength of this relationship. These findings were supported
both in Chapter 4 and by previous research looking at sexual aggressors of women and the
finding that rapists tended to be younger and in sexual relationships at the time of their
offence (e.g., Grubin, 1994; Milsom, Beech & Webster, 2003; Oliver, Beech, Fisher, &
Beckett, 2007; Proulx, Beauregard, Cusson, & Nicole, 2007)
Victim age was found to be positively related to the offender having any previous
convictions, a finding mirrored in McCann and Lussier (2008) who found that juveniles who
offended against adult victims (as opposed to peers) were more likely to continue to sexually
offend. Therefore, as the victim age increases the likelihood that the offender has a history of
offending is more probable.
Level of aggression exceeding that necessary to commit the offence implied that the
offender did not live alone, and was co-habiting. As it was found in Chapter 4, approximately
one quarter of the sexual aggressors lived with either a girlfriend or wife at the time of their
offence. This finding could be related to the victim acting as a substitute for another
108
significant woman in the offender’s life on which he is unable to take out his frustrations or
rage. He therefore finds a victim with which he can – the notion that the victim acted as a
vehicle for the offender (Fritzon & Ridgway, 2001). Potentially the expression of exaggerated
violence could also be related to an anger-led motivation, as expressed in the Angry pathway
found in Chapter 4, where the victim was used to expel their anger and frustrations, similar to
the Male Sex Drive is Uncontrollable and/or Dangerous World implicit theories (explored in
the next chapter) in which the offender’s loss of control is externalised to the victim, and the
perception of threat they pose to the offender.
5.7.1 Limitations
Any of the conclusions drawn from the above analyses need to be made with caution. Firstly,
the offender characteristics, with the exception of perpetrator age, were coded dichotomously.
This type of variable coding could also limit the more complex relationships between
variables to more simple ‘present’ or ‘absent’ relationships. Therefore, the analysis may not
have completely explored the full breadth of the possible relationships between the variables.
Secondly, as the sample is small, and the variables are non-parametric, the results may not be
indicative of a real effect, or they may be over-estimations of the magnitude of an association
between variables. The smaller the sample size, the lower the power of the test, and the
greater the likelihood of failing to reject a false null hypothesis (Hackshaw, 2008). The
sample would need to be 3 or 4 times larger to achieve enough power to be certain about the
findings.
Another constraint on the current findings is the limited number of possible contextual
variables that could be explored in analysis. While all four had some theoretical backing for
their influence on offending, they represent only a sample of the possible situational factors
that could affect the relationship between sexual offences and sexual offenders’
characteristics.
The current research is a starting point for the usage of mediation analysis and
bootstrapping techniques to look at the relationship between offence characteristics, offender
characteristics, and the possible intervening variables that may influence this relationship.
Future research needs to look at different contextual variables than those presented here, as
well as, the use of continuous or interval data within the analysis to allow for a richer
understanding of the interaction between the different levels of factors involved in predicting
offender characteristics from offence characteristics. Also, exploration of the effects of
109
concurrent multiple mediating variables is another possibility, as the use of alcohol and drugs
have also been linked to the location of victim contact and crime location (Chermack et al.,
2010).
110
Chapter 6: Implicit Theories as a Mediator between Crime Scene and Offender
Characteristics
6. Introduction
The notion that sexual offenders see the world and their victims in self-serving and distorted
manners is not a new one, although research has focused more on the actual content of their
cognitive distortions, than on the theory underpinning them. Ward (2000; Ward & Keenan,
1999) argues that implicit theories (ITs) are surface level measurements of offence-
supportive attitudes and cognitive distortions (Ward, Hudson, Johnston, & Marshall, 1997).
These ITs are comparable to “scientific theories” which can be used by the offender to
explain, predict and interpret the behaviours, thoughts, and beliefs of others (including their
victim), and about the world in general. ITs help to control and provide structure to a person’s
internal life and link the individual with their social environment and give meaning for the
events of their lives (Beech, Fisher, & Ward, 2005). Simply put, these perceptions of
ourselves, and our environment, will ultimately dictate how we react and behave in different
situations (Mischel & Shoda, 1995).
Looking at sexual offences specifically, Ward and Keenan (1999), Polaschek and
Ward (2002), Polaschek and Gannon (2004), and Beech et al. (2005; 2006) examine the
underlying causal theories surrounding the cognitive distortions expressed by samples of
child molesters, rapists, and sexual murderers. Overlapping ITs for all of these samples
include: Women/Children as Sexual Objects, Dangerous World, Uncontrollability, and
Entitlement 36. A further IT identified with regards to rapists of women is that Women are
Unknowable/Dangerous, which can be seen as a subset of the Dangerous World IT.
6.1 “Women/child as Sexual Object” Implicit Theory
The Women/children as Sexual Objects IT sees people, both adults and children, as being
motivated by their desire for pleasure, as sexual beings that desire and enjoy sexual contact.
Within this theory, children have the cognitive ability, and are capable of identifying their
needs and sexual preferences. They possess the knowledge and desire to make decisions
about sex with adults, and even initiate and develop strategies to achieve their sexual goals.
As a result of this, the offender views sexual activity as harmless and beneficial for the child
36
With specific regard to child molesters, Nature of Harm was a further identified IT. Nature of Harm holds that there are degrees of harm
(ranging from little or no harm to extreme damage) and that sexual activity in itself is actually beneficial and unlikely to harm a child. As
well, sex is seen as inherently beneficial, and any negative connation is a result of the stigmas put on it from society etc., rather than from
the actual act of sexual experience itself (Ward & Keenan, 1999).
111
(Ward & Keenan, 1999). According to this IT, women who are perceived to be constantly
open to sexual advances and do not deliberately deceive men, but their nonsexual behaviour
may be misinterpreted by men as having a sexual intention, even if the woman is unaware of
the unconscious messages she sends. According to this IT, women are seen as constantly
desiring sex, whether consensual, coerced or violent, and are fundamentally seen as sex
objects (Beech et al., 2005; Polaschek & Ward, 2002). According to this theory, offenders
justify their actions because by denying one’s sexual needs, urges and desires, can distort
and/or negatively influence a person’s functioning. The expression of sex is a healthy and
sacred experience that leaves people feeling secure, loved, and which is only harmful in
extreme circumstances (Ward & Keenan, 1999).
6.2 “Dangerous World” Implicit Theory
The Dangerous World IT is based on the belief that people are inherently untrustworthy and
rejecting, and will behave in a way that promotes their own interests and gains, regardless of
others, making the world a dangerous place. In order for the offender to achieve dominance
or control over others it is necessary to fight back first (Beech et al., 2005; Polaschek &
Ward, 2002; Ward & Keenan, 1999). If a person, or child, is seen as a threat or as trying to
dominate or hurt the offender, they need to be punished in order for the offender to re-assert
their dominance and retribution. Or, on the flip side of this theory, the offender believes
children will never exploit or reject them as the child is seen as more innocent, reliable, and
accepting, and therefore as more trustworthy, and able to provide an offender with the
potential to obtain love and be cared for (Ward & Keenan, 1999). In contrast, women are
likely to become the victim of sexual abuse as they are perceived as threatening to the
offender. The perceived malicious intention of others to dominate or hurt the offender is the
focus of this IT (Beech et al., 2005; Polaschek & Ward, 2002).
6.3 “Male Sex Drive is Uncontrollable” Implicit Theory
The next IT, Uncontrollability, or Male Sex Drive is Uncontrollable, sees the world as
uncontrollable and inevitable in its actions. A person is not able to substantially change
anything that happens within their life, including their sexual feelings, or their emotions, all
these just happen. Human nature is seen to be determined early on and is so ingrained that it
cannot be altered, therefore early pertinent experiences (i.e., loss of a parent; sexual abuse)
112
can leave the offender with deviant preferences which cannot be suppressed, managed or
controlled. As these urges are beyond their control (external to the offender), the offender
does not see himself as responsible for them, or his sexually abusive behaviour, but may
direct blame towards the object of their sexual desires (i.e., the victim) or other external
factors such as stress or alcohol/drug intoxication (Ward & Keenan, 1999). Women are seen
as having a major influence in a man’s inability to control his sex drive, and are often the
main variable in the loss of his control, especially if they deny what the offender deems as
‘reasonable’ sexual access (Beech et al., 2005).
6.4 “Entitlement” Implicit Theory
The Entitlement IT suppositions that as some people, namely men, are more important and
more superior to others, their superiority gives them the power and the right to put their needs
and desires over and ahead of those of others, often those of women. This illusion of
grandeur, often based in gender, class, or some other factor, entitles the offender to special
consideration, who’s desires, beliefs, and needs are above everyone else’s, and who’s victim
is there to meet these regardless of their own desires and/or beliefs (Ward & Keenan, 1999).
This IT emphasises the offender’s desires and beliefs as paramount. The victim is expected to
acknowledge and accept this fact and that their own desires and beliefs are secondary and
will be ignored (Beech et al., 2005; Polaschek & Ward, 2002).
6.5 “Women are Unknowable/Dangerous” Implicit Theory
The final IT, Women are Unknowable/Dangerous, sees women as entities that cannot be
understood by men as they are intrinsically different from them, either due to biology or
socialisation. Women are seen to be aware of these differences, and the fact that their own
desires and needs are not compatible with those of men, and therefore they do not explicitly
try to communicate them to men, but rather they present them in a hidden and covert
(deceitful) manner (Beech et al., 2005; Polaschek & Gannon, 2004; Polaschek & Ward,
2002).
These implicit theories determine the way the offender will interpret different ‘cues’
and behaviours from their victims, and influence how the offender disregards or ignores
possible alternative interpretations. Seemly innocuous behaviours such as a child requesting a
hug from a family member may be interpreted as that child having a sexual interest in the
113
person by the offender, rather than the fact the child may want reassurance or is simply
showing affection. Alternatively, the buying of a drink and dinner for a woman makes her
morally obligated to have sex with the offender. The different implicit theories have different
primary focuses such as the offender (entitlement and uncontrollability), the victim (sexual
objects and women unknowable/dangerous), and the world (dangerous world), although they
all have implications for each of these. These theories allow the offender to process and
predict the behaviour of (potential) victims, and allow the offender to justify their offending
against children (Ward & Keenan, 1999) and women (Beech et al., 2005; Polaschek &
Gannon, 2004; Polaschek & Ward, 2002).
6.6 Aim of the Chapter
The purpose of the present study is to look at the relationship between the above identified
ITs of sexual offenders, the characteristics of the offenders that hold different ITs, and the
potential effects these may have on the way the offence is committed. The inclusion of ITs
allows for a more robust understanding of the potential offender’s psychological makeup and
the underlying and internal psychological processes within an individual that results in their
behaviour(s). If relationships are found this may help in the investigative process during
witness and suspect interviews, as a means to identify attitudes that may be offence
supportive. As the different ITs are related to the schemas the offender holds about their
world and those they interact with, they may influence how they interpret and react during
their offence (Elliott, 2010) to situational factors such as potential interruptions, or victim
resistance. ITs can be viewed as forms of cognitive-affective units (CAUs) which are the
mental representations of the “individual’s construals, goals, expectations, beliefs, and
affects...” (Mischel, Shoda, Mendoza-Denton, 2002, p. 53). These CAUs are stable within the
personality and temperament of the individual, and evolve from the interaction and
experiences based on culture, subculture (Mendoza-Denton, Shoda, Ayduk, & Mischel,
1999), and social learning history of the person (Mischel et al., 2002). Yet, while they are
stable units of cognitive-affect, they will vary according to the accessibility, organisation, and
interrelationships between the CAUs, changing situations, and interactions between the
person and those they come into contact with (Mischel et al., 2002). For example, an offender
who holds the Dangerous World IT may interpret victim resistance as a further example of
the victim trying to hurt them and could therefore increase their own use of violence to
subdue the victim. Funder and Colvin (1991) hypothesise that those behaviours which are
114
the result of more internal sources (e.g., schemas) may be more consistent across situations,
such as crime scenes, and over time. The results of their study found exactly this. Across
three different situations, behaviour coded according to its psychological meaning (e.g.,
fearfulness), as opposed to more concrete or specific micro-level units, were highly consistent
– giving support to the behavioural consistency hypothesis of Offender Profiling. It could
also be hypothesised that these ITs would be related to different offender characteristics as
age and life experiences may shape the offender’s internal schemas. Attitudes towards
women and non-consensual sex can be influenced by and develop through the individual’s
attempt to understand their own experiences, and the attitudes of significant others in their
life (e.g., friends, family, and abusers) (Hanson & Morton-Bourgon, 2005). While, their
behaviour may not be exact across situations, the influence of and interrelationships of their
CAUs and personality will be (Funder & Colvin, 1991; Mischel et al., 2002).
Some of the current research on ITs has looked at a few offender characteristics and
offence characteristics. Beech et al. (2005) sorted their sample of 28 sexual murderers
according to most common ITs and found three main groups by IT: 1) Group 1, heavily
influenced by the co-occurrence of Dangerous World IT and Male Sex Drive is
Uncontrollable ITs were more sadistically driven; 2) while Group 2, dominated by the
Dangerous World IT only, were grievance driven; and 3) Group 3 were sexually driven,
consisting of only the Male Sex Drive is Uncontrollable IT. They also looked at some offence
characteristic difference between the three groups, supporting the notion that some ITs may
be related to particular crime scene behaviours. Group 1 members reported that they had a
prior intention to kill their victims, and that their offence was a means to carry out sadistic
thoughts and fantasies against strangers, often sexually interfering with the body post-
mortem. Moreover, they had a significantly higher history of reported violence against
women. Resentment and anger towards women were the dominate motivations for Group 2.
They often offend against women they knew, and they had the highest level of convictions
for nonsexual and/or nonviolent offence out of the three groups. Group 3’s offences were
motivated by their thoughts and fantasies just prior to their offence of the prospect of having
sex. They targeted adult women, the killing of the victim to avoid detection, or to keep their
victim quiet.
Beech et al.’s grouping of offenders by both ITs, offender, and offence demographics
shows that offenders can be differentiated by their ITs. Furthermore, these may correspond to
different offence behaviour and offender characteristics which further develop the typology
of sexual aggressors. That said, the level of statistical analysis done was at a basic level; the
115
percentages and frequencies of the occurrence of the ITs and different characteristics were
calculated, alongside simple chi squares comparisons between the groups. The present
chapter will aim to expand on what Beech et al. found and look to explore whether there are
more substantial and significant relationships between the offender characteristics, with
regards to the different IT, and offence characteristics. This will be done through mediation
analysis.
6.7 Method
6.7.1 Sample
committed and were convicted of at least one sexual offence against an adult (16 years of age
or older) female victim, and who were taking part in the Core SOTP between 1998 and 2002.
6.7.2 Content Analysis
The available transcripts and offence details for 94 of the original sample of 102 sexual
aggressors were coded according to the content dictionary (Appendix E). The transcripts
were of semi-structured interviews conducted with the offenders that asked about their
experiences of treatment, during which they were asked about their offence(s). The questions
were open-ended, and the first section of the interview questioned who the offender had
offended against, their relationship to the victim, and what had led to their offence. Questions
surrounding the motivation of the offence were asked by the interviewer to determine if the
offence was related to sex or to anger. Other prompted questions were based upon the
offenders’ answers.
The content dictionary was developed using the information contained in the articles
by Ward and Keenan (1999), Polaschek and Ward (2002), Polaschek and Gannon (2004), and
Beech et al. (2005) regarding the identified implicit theories. The transcripts were
individually coded by the primary researcher using the five ITs as coding categories. If an IT
was evident from the transcript or offence details, it was recorded. The IT that was most
dominant in the data was coded as the ‘primary’ IT and used in the analysis. This was done as
the ITs are not mutually exclusive of one another and there was often more than one IT
evident in the offenders’ transcripts and offence details (as was the case in Beech et al.,
2005). The primary IT was thought to best represent the influencing offence-supportive
116
cognitions activated at the time of the offence. The name of the IT was recorded for each of
the available 94 cases. A random subset of 20 anonymised transcripts was coded by a second
independent researcher who had no knowledge of the first coder’s results. The inter-rater
reliability of whether an IT was present in each account was assessed using Cohen’s kappa.
The overall agreement of the categorisation of the cases was high (k = 0.80; p < 0.001),
indicating a substantial level of agreement (Landis & Koch, 1977).
See Table 6.1 for the coding of the offence characteristics (X), the offender characteristics
(Y), and for the coding of the mediating variables (M), which in this chapter are the ITs. Each
of the offence and mediating variables were coded according to whether they could be
identified from the actual offence, crime scene, victim, and the offence details. The offender
characteristics were coded using the information available through the functional analysis
questionnaire and the offender files.
37
See Chapter 3 for more detailed description of the coded variables.
117
Table 6.1
Coding of Offence Characteristics (X), Offender Characteristics (Y), and Potential Mediators
(Y; Implicit Theories)
Offence Characteristics Coding
99=unknown
1=minor injuries
3=major injuries
99=unknown
2=rape
Victim Age Interval
16 to 86 years
1=used
99=unknown
Sexual Penetration 0=none
Offender Characteristics Coding
Perpetrator Age Interval
14 to 57 years
Relationship Status 0=no
1=yes
Any Previous Convictions 0=no
1=yes
Lives Alone 0=‘other’
1=alone
Employed 0=‘other’
1=employed/student
Potential Mediators: Implicit Theories Coding
Women Unknowable/Dangerous 0=not indicated
1=indicated
Women as Sex Object 0=not indicated
1=indicated
Male Sex Drive is Uncontrollable 0=not indicated
1=indicated
Dangerous World 0=not indicated
1=indicated
Entitlement 0=not indicated
1=indicated
118
For some of the analyses there was incomplete information for all of the offenders, as the
information was not coded for, was missing from the hard file of the Functional Analysis
questionnaire, could not be ascertained from the transcripts of the semi-structured interview,
or from the offence details. For the perpetrator age analyses, information was available for 67
of the 94 offenders, except for the perpetrator age and level of injury analysis, which had a
sample of 45, as there were missing variables for level of injury within the dataset.
Relationship status, any previous convictions, lives alone, and employment status analysis
with level of injury had a sample of 56 due to missing variables. The rest of the analyses had
a sample of 94.
6.8 Results
6.8.1 Occurrence of Implicit Theories
Four of the five ITs were present in the current sample. Dangerous World was the most
frequently identified primary IT (40%), followed by Male Sex Drive Uncontrollable (31%),
then Women as Sex Objects (22%), and lastly, Entitlement (6%). Women are
Unknown/Dangerous was not indicated within the sample as a primary IT.
correlations between variables was above 0.40, the variance of inflation (VIF) ranged from
indicate a low likelihood of issues related to multicollinearity affecting the regression model
(Field, 2009).
6.8.3 Mediation (Paths a and b), Indirect (ab), and Total Effects (c)
characteristics. The figures for these analyses can be found in Appendix F, along with the
tables of the Indirect and Direct Effects. Those that were significant are presented below.
119
Overall, there were no fully or partially mediated relationships between the various
offender characteristics, implicit theories (mediators) and offence characteristics. Concerning
the indirect effects 38, the only relationship was found between sexual penetration behaviours
and the offender living alone, when an entitlement IT was held by the offender as indicated
by the confidence interval for this relationship, which did not include zero, 0.166 to 1.219
(see Table 6.2). In the presence of the Entitlement IT, if the offender had committed the more
deviant sexually penetrative behaviours then they were more likely to be living alone at the
time of the offence.
38
The indirect effects of ab are the combined effects of paths a and b and their relationship with Y (Offender characteristics), independent of
the direct effects of X (offence characteristics) on Y (Hayes, 2009; Preacher & Hayes, 2008).
120
Table 6.2
Indirect Effects of Sexual Penetration on Lives Alone through Implicit Theories (ab paths)
Bootstrapping

CI
Dangerous World -0.032 0.088 -0.202 0.051 -0.278 0.026 -0.254 0.030
Women as Sex Objects -0.001 0.090 -0.065 0.089 -0.083 0.071 -0.094 0.061
Male Sex Drive
Uncontrollable -0.018 0.073 -0.148 0.081 -0.183 0.059 -0.160 0.069
Entitlement 0.539 0.235 0.166 1.080 0.178 1.131 0.202 1.219
Significant total effect 39 relationships were found for perpetrator age, level of injuries and
offence outcome (Table 6.3); relationship status and offence outcome (Table 6.4); and
previous convictions, level of injuries and victim age (Table 6.5). The analysis indicates that
the relationship between perpetrator age and level of injury was negative, therefore, as the
level of injury increased it was more likely that the perpetrator’s age decreased as well. The
outcomes of the offence (rape or murder) also had a significant relationship with perpetrator
age, with rape more likely indicating an older offender and murder a younger offender. This
last finding is in line with the finding of Chapter 4 and 5 concerning the offenders’ ages and
offence outcome.
It was more likely that the offender was in a sexual relationship at the time of offence
if the sexual offence was a rape, whereas, if the offence ended with the death of the victim
then it was a greater possibility that the offender was not in a sexual relationship at the time
of the offence. A positive relationship was found between the level of injury and previous
convictions, indicating that as the level of injuries sustained by the victim increased in
severity so did the likelihood that the offender had a previous conviction. There was a
positive relationship between victim age and the offender having a previous conviction
indicating that as the victim age increased so did the likelihood that the offender had previous
39
The total effect (c) is the sum of the direct effect of offence characteristics (X) on offender characteristics (Y; c’) and the indirect effect of
offence characteristics on offender characteristics through the implicit theories (X on Y through M; ab) (Hayes, 2009; Preacher & Hayes,
2008).
121
convictions. These latter two relationships were also in line with findings from Chapter 5. No
direct effects were found between the offender living alone or being employed at the time of
the offence and any of the offence characteristics.
Table 6.3
Product of
Coefficients
B SE t p
Offence Outcome 4.321 1.850 2.335 0.023
Victim Age -0.015 0.051 -0.286 0.776
Weapon Used -2.555 2.235 -1.143 0.257
Table 6.4
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.894 0.451 1.982 0.047 3.093
Victim Age -0.008 0.011 -0.686 0.493 0.471
Weapon Used -0.435 0.527 -0.827 0.408 0.684
122
Table 6.5
Total Effects of Offence Variables on Previous Convictions (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.036 0.533 0.068 0.946 0.005
Victim Age 0.052 0.024 2.163 0.031 4.679
Weapon Used -0.065 0.633 -0.102 0.919 0.010
6.8.4 Direct Effect of Offence Characteristics (X) on Offender Characteristics (Y) (c’) 40
There were two groups of significant c’ paths for perpetrator age: 1) level of injury and 2)
offence outcome. The relationship between perpetrator age and level of injury was negative,
indicating that as the offender’s age increased the level of injury to the victim decreased. The
relationship between offender age and offence outcome was positive, indicating as the
offender’s age increased so did the likelihood that the offence was a rape, as opposed to a
sexual murder.
There was a significant c’ path between relationship status and offence outcome,
although these were bordering significance (p=0.05). Again, as with the total effect, and the
findings in Chapter 4 and 5, this indicates that a rapist was more likely to be in a relationship
at the time of their offence, than a sexual murderer.
There were two groups of significant c’ paths for the offender having any previous
convictions: 1) level of injury and 2) offence outcome. These findings indicate that as the
level of injury increased so did the likelihood that the offender had previous convictions, and
that the offence ended in a rape only. There were no significant c’ paths for the offender
living alone or being employed at the time of their offence.
40
Path c’ is the direct effect of offence characteristics (X) on offender characteristics (Y) independent of the effects of implicit theories (M)
(Hayes, 2009; Preacher & Hayes, 2008).
123
6.9 Discussion
The aim of the chapter was to test for mediating relationships between offender
characteristics, offence characteristics and possible implicit theories the offenders have with
regards to their victim and the world in general. There was no indication of any partial or full
mediation between the three groups of variables, suggesting no associated between the
offenders’ ITs, their offence characteristics, and their own characteristics. Even though there
were no direct links between the offenders’ ITs, offence details, and characteristics, the ITs
were still identified within the current sample, as they have been in previous samples of
rapists and sexual murderers (e.g., Beech et al., 2005; 2006). The offender’s IT will influence
how they interpret other people’s actions, thoughts, and beliefs, and how they make
predictions about the world. The offender will seek out environments and others that support
their ITs (Beech et al., 2005). Despite the fact that there were no direct links found, this does
not suggest or mean that the ITs do not influence the offenders, their actions, or their
interpretation of their victim’s thoughts and behaviours as both direct and total effect
relationships were found.
Significant direct effect relationships (paths c’) and total effect relationships (path c)
were found between offender age and level of injury and offence outcome (rape/murder);
relationship status and offence outcome; and previous convictions and level of injury and
victim age. To recap, path c in the mediation analysis represents the total effect of paths ab
plus the direct effect of path c’, and thereby it encompasses the influence of all involved
variables on one another. Path c’ is the “effect of X on Y that is independent of the pathway
through M” (predicting Y from X) (Hayes, 2009, p. 409), meaning it represents the direct
relationship between Y and X. What the finding of both of these relationships indicates is that
while there are direct relationships between some of the offence and offender characteristics,
the proposed mediators, the ITs, are not necessary for this relationship to exist. Despite this
fact, the ITs do have some influence in the strength of these relationships, whether it is by
strengthening or weakening them. Even though the focus of the study was in the pursuit of
mediating relationships, it is evident from the findings that that the relation between some of
the offence and offender characteristics and ITs is more likely that of a moderated one (Baron
& Kenny, 1986).
According to Baron and Kenny (1986), moderation implies that the relationship
between two variables, the predictor and dependent variable, changes in direction and/or
strength as a function of a third variable, the moderator. Moderation does not look at causal
124
relationships, but under what conditions does a variable have an effect on the outcome– under
what conditions of B (moderator; IT) is A (predictor; offence characteristic) significantly
associated with C (outcome; offender characteristics) (Baron & Kenny, 1986). It may be that
the offender’s implicit theories are not causal factors in determining their offences and who
they are, but they may play a significant role in altering the strength of the associations
between who they are and how they offend.
Looking at the differences between the coefficients for the significant c and c’ paths
for the relationship between perpetrator age and level of injury sustained by the victim, the
inclusion of the ITs weakened the direct relationship between these variables. This was also
the case between perpetrator age and offence outcome. The opposite effect was found for
relationship status and offence outcome, previous convictions and level of injuries, with the
inclusion of the ITs strengthening the relationship between the predictor and dependent
variables. Although, for previous convictions and victim age the coefficient was the same for
both paths c and c’, meaning there was no change in the relationship with the inclusion of the
third variable, ITs.
Within the current dataset it was found that as the level of victim injury increased the
offender age decreased. This indicates that younger offenders were more likely to cause more
injuries to their victims. Combined with the current finding of a positive relationship between
offence outcome and offender age (older offender more likely outcome of rape, not sexual
murder), this is similar to the findings of Harry, Pierson, and Kuznetsov (1993). They found
aggressive younger offenders tended to attack older victims, with the increase in violence and
injury leading to more victim deaths. This could be due to older offenders using less injurious
practices, or the utilisation of more strategic forms of coercion (e.g., verbal threats) within the
offence, or possibly as a result of increasing knowledge of offending (i.e., practice of
offending; multiple attempts) as the offender ages. Groth (1977) also found that victims of
the same age or those significantly older than the offender were more likely to have a weapon
used against them, increasing the potential for resulting victim injury. This finding conflicts
with the positive relationship between offender age and victim injury found in McDermott
and Hindelang (1981) and Felson and Krohn (1990); with older offenders more likely to use
non-strategic violence 41 and weapons, resulting in higher levels of victim injury, and possible
death. They postulate this could be a result of younger offenders offending for more sexually
41
Strategic violence is only what is necessary to complete the offence, whereas, non-strategic violence, threatened or overt, is violence
which is beyond what is necessary or needed in order for the offender to commit their offence (Felson & Krohn, 1990).
125
motivated reasons and therefore, being less likely to harm their victim unless it is necessary
to complete their crime, facilitate escape, or avoid prosecution (Felson & Krohn, 1990).
These findings could also be linked to the offender’s attitudes towards women, even if
not found directly in the current study. Depending on the engrained IT theories, offender age,
and/or experiences, if the offender views women as sex objects or as a threat in some
perceived form, then this may disinhibit the use of violence and increase the likelihood of a
sexual assault occurring, as they re-interpret situational and victim behaviours to match their
internal schemas. Research by Malamuth (1986) suggests that sexual arousal to rape and
various attitudes towards women may decrease inhibition in some men, leading to more
coercive sexual behaviours. It could be hypothesised that as the offender ages and gains more
experience both generally and in his offending, his practices and attitudes may evolve and
change. This could lead to more strategic violence being used as he learns ‘better’ more
efficient ways to effectively commit his offences and control his victims.
Older victims and increased victim injury both indicated a greater likelihood of the
offender having previous convictions. This finding is in line with Francis and Soothill (2000),
and Rice, Harris, and Quinsey (1990). They found having previous violent sexual offences
often indicated that the offender was more likely to commit further, often more serious,
violent offences (i.e., homicide). Furthermore, offenders who offended against adult (older)
victims, use more violence and inflict more injury to their victim (Kuznetsov, Pierson, &
Harry, 1992). These offenders were also those that had higher degrees of non-sexual criminal
histories (Harry et al., 1993). This idea of the escalation of violence and offending behaviour
is not novel and is briefly described in Chapter 4).
The positive relationship between victim injury and previous convictions could
indicate that the offender may be using his victim as more of a “vehicle” for his expression of
anger, resentment, or frustrations towards the world and system he has been involved in
(Canter & Heritage, 1990). The offender’s victim can act as a surrogate for other people (i.e.,
girlfriend, wife, mother) and/or experiences (i.e., prison, criminal justice system) in his life
that he has perceived as having harmed or taken advantage of him, or who he perceives as
threatening to him. The offender uses excessive violence, verbally, physically, and sexually
(e.g., anal penetration), against the victim because they are seen as a general target upon
which the offender can vent his frustrations. They do not represent a specific person or
object, but some generalised ‘other’ – consistent with the Dangerous World IT. It could be
postulated that offenders with previous convictions and therefore bad experiences with the
criminal justice system may express and act out their anger towards this system and their
126
experiences within the system on their victim. This supports, in part, the escalation research
showing that re-offending sexual offenders escalate in their sexual offending. Often at a
higher frequency, inflicting more serious injuries, and committing more sadistic and callous
acts (e.g., foreign object insertion) during their latter offences (Hazelwood, Reboussin, &
Warren, 1989; Warren et al., 1999), and in the Warren et al. study the Increasers also sexually
assault significantly older victims (40 years or older) even when offender age was controlled
for.
As was found in Chapter 4, an offence outcome of rape was more likely for an
offender who was currently in a sexual relationship at the time of the offence, often with
which they were dissatisfied. The offence may be a way for the offender to gain sexual
gratification, or to vent his general and/or sexual frustrations, or to ‘punish’ his partner for
any inadequacies he may perceive are her fault or for which he blames her. This may be
especially true if he holds the Women as Sex Objects, Male Sex Drive is Uncontrollable
and/or Dangerous World ITs. Or this finding may be linked with the fact that sexual
murderers in general have been found to be more socially isolated (Milsom, Beech, &
Webster, 2003), have difficulty maintaining social relationships because of low self-esteem,
deficient social skills and feelings of inadequacies (Beauregard, Stone, Proulx, & Michaud,
2008; Groth & Birnbaum, 1979; Pardue & Arrigo, 2008), and are therefore less likely to be in
a sexual relationship at any given time.
6.9.1 Limitations
The lack of any mediating relationship within the current chapter could be a result of how the
variables were coded. While the inter-rater reliability was high, and the content dictionary
was based on the descriptions provided in various implicit theory articles, it could be that the
concepts were not properly captured from the dataset when coding for the different implicit
theories. This may be due to singular coding of only the primary IT. If the ITs interact to
influence behaviour then by only coding for the ‘primary’ indication may suggest that
meaning is lost with regards to the underlying mechanism behind the offence behaviours.
As the ITs were studied using interviews and file review the present research uses
historic and retrospective data, which could be an influencing factor, as often the interviews
were conducted some time after the offence. How the offender viewed their reasons for
offending and the indications of existing ITs at the time of interview may not represent those,
which were actually activated or present at the time of the offence. As well, the interviews
127
were not conducted for the purpose of the current study, so the questions asked and responses
given may not entirely encapsulate the necessary information to assess which IT was actually
present. Although, studies utilising questionnaires, interviews, and file-based measures have
been successful in previous cognitive distortion and IT research (e.g., Beech et al., 2005,
2006; Polaschek & Gannon, 2004; Pollock & Hashmall, 1991).
The reliance on self-report in assessing the presence of ITs is also a potential
limitation as this methodology relies on the assumption the offender could readily identify
their motivations and reasons for offending and are able to express these during the interview.
It may also be the case that the offenders, who were all on or had completed a sex offender
treatment programme, were downplaying the extent of their sexist attitudes to show that they
are capable of rehabilitation and reform. If this is the case, then it may be argued that the
occurrence and significance of the ITs are higher than they were indicated to be in the current
sample.
While the offender may hold a certain implicit theory, this does not have an apparent
causal effect on their behaviour during the commission of their offence for the offence
variables chosen, and does not directly link back to their own individual characteristics. The
offence process may be influenced by the ITs the offender holds, as is more evident in the
current findings by the fact that both the direct and total effects for a few of the variables
were both significant. Yet, it is not clear whether the ITs initiate the sexual offending
behaviour or if they act to maintain sexual offending behaviour once it has started. There is
also for the potential of situational and contextual factors, which interact with the ITs, to
influence the offence process as well.
While the current study did not find support for causal relationships concerning the
ITs, offender characteristics, and offence characteristics, there is no question that the ITs held
by the offender will influence how they perceive and interact with their surroundings, and the
people they come into contact with, both potential victims, and in more general everyday life.
128
Chapter 7: Motivations as a Mediator of the Interaction between Crime Scene and
7. Introduction
Historically, the motivation often assumed to be behind a sexual offence has been that of
sexual gratification, the sexual thrill of the offence, and/or the sexual intoxication that
accompanies forcing sexual contact. However, other non-sexual motivations, such as the
expression and/or displacement of anger and the exertion of power and control over another
human being, have also been given support (Groth & Birnbaum, 1979; Pardue & Arrigo,
2008). It is evident in the use of physical violence displayed in a sexual offence that sexual
offenders are not only and are not all driven by sexual impulses. It is acknowledge that sexual
offences are committed for different reasons and that within any sexual offence there will be
evidence of different motivations (Fisher & Beech, 2007), with some form of power and
control being entwined throughout all of them; the most dangerous being that of a sadistic
nature (Robertiello & Terry, 2007). Various typologies of the motivations for sexual
offending have been developed to try to provide a more cohesive understanding of these
offences, and to help guide clinical judgement (i.e., Beech, Fisher, & Ward, 2005; Fisher &
Beech, 2005; Groth, Burgess, & Holmstrom, 1977; Keppel & Walter, 1999; Knight &
Prentky, 1990; Proulx & Beauregard, 2009a, b). Many, if not all, of these, and other
classifications of sexual offenders allude to or infer some type of motivation in their use of
terms such as ‘sexual/ly’, ‘control’, ‘sadistic’, ‘power’, ‘anger/aggressive’, etc., as well as,
make inferences about the motivations surrounding why certain behaviours were expressed
during a rape or sexual murder. Yet, the determination of motivation is not always
straightforward or easy to imply from the available offence information. Nevertheless, it is
too important an aspect of any behaviour to be ignored or left out – motivation is necessary
for any sexual offending to occur (Palmer, DiBari, & Wright, 1999). It is an imperative piece
of the puzzle into answering the question ‘Why’ – why an offender chose to commit his
crime, why he chose his victim, and why he committed the crime in the manner he did.
Generally, the study of sexual offenders has been for the purposes of clinical practice,
risk assessment, and for the development of therapeutic interventions. Yet, most of the
typologies developed with regards to sexual offenders could be of use for offender profiling
as well (Beauregard, 2010). The different typologies (i.e., the FBI typology; Groth &
Birnbaum, 1979; Knight & Prentky, 1987) focus on the offending process (i.e., how the
129
offender may have approached their victim; victim choice; possible reasons behind certain
sexual aspects of the offence), and can help guide the search for unknown suspects. Offender
profiling is concerned with building a portrait of an unknown offender using the available
crime scene and victim characteristics, the inclusion of potential motivational aspects evident
in these will provide a more detailed picture of the possible offender’s psychological
constitute, leading to more accurate inferences made about the possible suspects, and the
possibility of more concise elicitation and prioritisation of these potential offenders. Four
main types of motivations are evident in almost all typologies: opportunistic, compensatory,
angry, and sadistic, and each of these will now be described.
7.1 Sexually Motivated: Opportunistic or Compensatory

The sexually motivated offender is preoccupied by sex, often has a feeling of powerlessness
about their lives, and is driven by sexual fantasies about their victims (Beech, Fisher, &
Ward, 2005; Keppel & Walther, 1999). Two different types have been identified. The
sexually opportunistic offender will commit his sexual offence often during the course of
another crime, is influenced by situational and contextual factors (Knight, Warren,
Reboussin, & Soley, 1998), and often will have used alcohol before his crime (Proulx &
Beauregard, 2009a). In contrast, the more sexually compensatory offender will explicitly plan
their sexual offence (Vettor, Beech & Woodhams, in press). With regards to both, the use of
strength, authority and control over their victims is a means for the offender to accommodate
their feelings of inadequacy and to affirm their masculinity (Groth & Birnbaum, 1979;
Knight, 1999) or to compensate for their perceived lack of positive self-image (Pardue &
Arrigo, 2008). These opportunistic or compensatory sexual offenders will most likely have
previous criminal convictions, be less likely to have steady employment and have more idle
lives in general, indicating a generalised conflict with society and an antisocial lifestyle. They
may also have a large number of sexual partners, relating to their need for intimacy and
continual reassurance of their masculinity (Pardue & Arrigo, 2008).
Both of these sexually motivated offenders will use minimal, more strategic violence.
Victim humiliation or mutilation, and severe injury are uncommon in their offences (Groth &
Birnbaum, 1979; Proulx & Beauregard, 2009a), resulting in lower victim mortality (Groth &
Birnbaum, 1979; Vettor et al., in press). However, victim death may result from ‘overkill’ as
the offender attempts to re-gain his sense of control over the victim if the victim does not
respond in a manner that fits the offender’s preconceived fantasies (Keppel & Walter, 1999)
or as a way to avoid detection (Beech et al., 2005; Clarke & Carter, 2000).
130
The behaviour of these sexually motivated offenders reflect similar behaviours of the
rapists in some typology systems, such as the power rapists (Groth & Birnbaum, 1979), the
power-assertive or power-reassurance offender (Keppel & Walter, 1999), the sexual
opportunistic or non-sadist rapist (Knight & Prentky, 1990), the sexually triggered/aggressive
control offender (Clarke & Carter, 2000) and the sexually motivated offender (Beech et al.,
2005; Beech, Ward, & Fisher, 2006; Fisher & Beech, 2005).
7.2 Anger/Aggressively Motivated
According to Groth and Birnbaum (1979), rape is always an aggressive act. For those sexual
offenders whose motivation is fuelled by anger or by aggression, the sexual aspect of the
offence is not for their sexual gratification, but is an expression of frustration, resentment,
and/or rage, and is a way to punish and humiliate their victims (Myers, Husted, Safarik, &
O’Toole, 2006; Pardue & Arrigo, 2008).
The level of aggression for these offenders ranges from verbal abuse to extreme
violence (Knight & Prentky, 1987; McCabe & Wauchope, 2005), but is usually far more
excessive than what is needed to control the victim (Pardue & Arrigo, 2008). Often these
anger motivated sexual offences are spontaneous and unplanned, shorter in length, and are
fuelled by a build up of frustration and/or induced by life circumstances (Groth & Birnbaum,
1979; Groth, Burgess, & Holmstrom, 1977). Victim death as a result from their attack could
be linked to the offender’s desire to physically hurt and punish their victim, often resulting in
extreme episodes of violence and anger (Beech et al., 2005). The angry sexual offender is
driven by their need to achieve dominance and control over their victim as they seek
vengeance against a perceived wrong brought against them (Beech et al., 2005).
These anger motivated offenders reflect similar behaviours reported for angry rapists
(Groth & Birnbaum, 1979), the anger-retaliatory offender (Keppel & Walter, 1999), the
pervasively angry or vindictive offenders (Knight & Prentky, 1990), the sexually
triggered/aggressive dyscontrol offender (Clarke & Carter, 2000) and the grievance/violent
motivated offender (Beech et al., 2005; Beech, Ward, & Fisher, 2006; Fisher & Beech, 2005),
that form part of published typologies.
131
7.3 Sadistically Motivated
Myers et al. (2006) argue the primary motivation behind the offending behaviours of all
sexual offenders is that of sadistic pleasure. Any other motivation (i.e., sexual gratification,
anger, control or power) has secondary purposes – to either increase or heighten the sexual
arousal, or the practical use of managing the victim so the offence can be carried out. While
this may be true for a subsection of sexual offenders, as evidenced above, there are other
primary motivations for sexual offending besides that of sadistic pleasure. Sadistic sexual
offenders are represented by their fusion of sexual and violent motivations and the enjoyment
of humiliating and distressing their victim (Beech et al., 2006). Sadistic sexual offenders use
the physical and psychological suffering they inflict upon their victim to fuel their sexual
excitement and arousal (Douglas, Burgess, Burgess, & Ressler, 2006). Their victim is seen as
an inanimate object, towards which the offender has no empathy or concern (Canter, Bennell,
Alison, & Reddy, 2003). They will often commit several types of torture (i.e., bondage,
foreign object insertion, sexual mutilation) and degradation (i.e., hair cutting, cigarette
burning, sex with a corpse) against their victim and feel no remorse for their actions (Canter
et al., 2003; Pardue & Arrigo, 2008; Robertiello & Terry, 2007).
The sadistic sexual offender carefully plans his offences, often in the form of
elaborate sexual fantasies (Deu & Edelmann, 1997), and takes preventive measures against
being discovered, frequently murdering their victim as a precautionary measure (Groth &
Birnbaum, 1979), the threat of which is meant to cause terror for the victim. He will often
prolong contact with the victim, with the assault lasting hours or even days. He will have
elaborate deviant and violent sexual fantasies which play a significant role in their offences
(Deu & Edelmann, 1997; Pardue & Arrigo, 2008), often acting as offence-scripts (Pithers, et
al., 1988; Ward & Hudson, 2000). The sadistic sexual offender is, more often than not,
capable of functioning in society and recognising social constraints, but his narcissistic and
egocentric personalities override any social norms and morals (Dietz, Hazelwood, & Warren,
1990).
These sadistically motivated offenders share similar behaviours of the following
categories from typologies: anger-excitation sexual offender (Groth & Birnbaum, 1979;
Keppel & Walter, 1999), the overt-sadistic rapist (Knight & Prentky, 1990), the sexually
motivated offender (Clarke & Carter, 2000) and the sexually sadistic offender (Beech et al.,
2005; Beech, Ward, & Fisher, 2006; Fisher & Beech, 2005).
132
7.4 Aim of the Chapter
The purpose of the present study is to look at the relationship between the motivations of
sexual aggressors (sexual, sadistic, opportunistic or compensatory), the characteristics of
these offenders, and the characteristics of their offences. The inclusion of motivations allows
for the understanding of why the offence occurred, and of the offender who committed it.
Motivations link with the implicit theories explored in the previous chapter, as well as, being
integrated as part of the cognitive affective units of the CAPS model, as a part of the
psychological makeup of the offender and the underlying processes that lead to their
offending behaviour. If the psychological features of a situation can be determined then this
information can be used to predict future behaviour in a broader range of situations that
contain similar psychological features (Mischel & Shoda, 1995). These psychological
features will interact with the situation with slight variations occurring, yet the psychological
meaning of them will remain consistent, determined by the stable personality aspects of the
individual (Mischel & Shoda, 1995).
While, motivations themselves may not be explicitly found at a crime scene, if certain
crime scene aspects (e.g., level of injury; level of aggression exhibited; offence outcome) can
be linked to motivations, then these can be used to infer the motives of the offender at the
time of the offence, and therefore used in the development and the understanding of the
offender and their corresponding profile. If it is found that the level of violence used is
greater than that necessary to control their victim and complete their offence, it could be
hypothesised that the offender was motivated by anger, also suggesting that the offender will
have an extensive history of previous criminal convictions, comprising of sexual, violent, and
non-sexual/non-violent offences (Beech et al., 2005; Beech et al., 2006; Proulx &
Beauregard, 2009a,b). Similarly, sexual offences committed during other offences, such as
burglaries, with minimal force or injury to the victim, or the lack of weapon use, may indicate
an opportunistic or compensatory offender (Robertiello & Taylor, 2007), who have been
found to more often be unemployed, or hold inconsistent employment status (Proulx &
Beauregard, 2009a,b; Vettor et al., in press). The compensatory offenders are also more
likely to be in a sexual relationship as they are in constant need of reassurance and intimacy
(Pardue & Arrigo, 2008). These are just a few examples of motivations of the offender being
linked with various crime scene characteristics in the literature, which in turn can be used in
the inference of offender characteristics from the crime, and in the development of an
offender profile.
133
The other area that motivations may provide assistance in the criminal justice process is
in the punishment and treatment of sexual offenders once caught (Palmer et al., 1999). As
mentioned previously, certain aspect of sex offender treatment programs target the beliefs
and offence supportive cognitions that sustain the sexual offending behaviours of the
offenders. Part of effectively targeting these, is identifying the motivations behind the offence
and challenging them (Beech, Oliver, Fisher, & Beckett, 2005). As Robertiello and Terry
(2007) point out, an important aspect of reducing sexual recidivism is by identifying the
characteristics and motivations of the offender’s offending behaviours, and addressing these
accordingly. Chapter 4 of the current thesis, research by Proulx and Beauregard (2009a, b),
Beech et al. (2005; 2006), as well as others (i.e., Knight, 1999; Polaschek, Hudson, Ward, &
Siegert, 2001), have found various positive links between the various motivations and
offender and offence characteristics. The current study explores the question of motivation
specifically looking at the relationship between identified motivations, specific offender and
offence characteristics and testing for actual links between these with the goal that they may
be valuable in the production of an offender profile, the generation of suspect lists and
prioritisation, as well as potential links with treatment.
7.5 Method
7.5.1 Sample 42
committed and were convicted of at least one sexual offence against an adult (16 years of age
or older) female victim, and who were taking part in the Core SOTP between 1998 and 2002.
7.5.2 Content Analysis
The available transcripts and offence details for 94 of the original sample of 102 sexual
aggressors were coded according to the descriptions provided in the literature for Sadistic,
Angry, Sexually Compensatory, and Sexually Opportunistic offenders (e.g., Beech et al.,
2005; Beech et al., 2006; Clarke & Carter, 2000; Dietz et al., 1990; Fisher & Beech, 2005;
Groth & Birnbaum, 1979; Keppel & Walter, 1999; Knight & Prentky, 1990; Pardue &
Arrigo, 2008) (Appendix G). The transcripts were individually coded by the primary
42
As was the case in Chapter 6, there was incomplete information for some of the offenders. See Chapter 6 section 6.7.3 for a description of
the incomplete data.
134
researcher using the four motivations as coding categories. The motivation that was most
dominant in the data was coded as the ‘primary’ motivation and used in the analysis. This
was done as the motivations are not mutually exclusive of one another and there was often
more than one motivation evident in the offenders’ transcripts and offence details.
See Table 7.1 for the coding of the offence characteristics (X), and Table 7.2 for the coding of
the offender characteristics (Y). Each of the offence variables were coded according to
whether they could be identified from the actual offence, crime scene, and victim, as well as
the offence details. The offender characteristics were coded for through the use of the
information available through the functional analysis questionnaire and the offender files.
43
See Chapter 3 for a more detailed description of the coded variables.
135
Table 7.1
Coding of Offence Characteristics (X), Offender Characteristics (Y), and Potential
Mediators (M)
1=minor injuries
3=major injuries
4=unknown
2=rape
Victim Age Interval; 16 to 86 years of age
1=used
Sexual Penetration 0=none/missing
Offender Characteristics (Y) Coding
Perpetrator Age Interval; 14 to 57 years of age
Relationship Status (in a sexual relationship 0=no
at time of offence) 1=yes
Any previous Convictions 0=no
1=yes
Lives Alone 0=’other’
1=alone
Employed 0=’other’
1=employed/student
Potential Mediators (M) Coding
Angry Motivation 0=not indicated
1=indicated
Sadistic Motivation 0=not indicated
1=indicated
Sexually Opportunistic 0=not indicated
1=indicated
Sexually Compensatory 0=not indicated
1=indicated
For some of the analyses there was incomplete information for all of the offenders, as the
information was not coded for or was missing from the hard file of the Functional Analysis
questionnaire, or could not be ascertained from the transcripts of the semi-structured
interview, or from the offence details. For the perpetrator age analyses, information was
136
available for 70 of the 90 offenders, except for the perpetrator age and level of injury
analysis, which had a sample of 45, as there were missing variables for level of injury within
the dataset. Relationship status, any previous convictions, lives alone, and employment status
analysis with level of injury had a sample of 56 due to missing variables. The rest of the
analyses had a sample of 90.
7.6 Results
7.6.1 Occurrence of the Motivations
Anger was the most frequently identified motivation (40%), followed by Sadistic motivation
(23%), then Sexually Opportunistic (20%), and lastly, Sexually Compensatory (17%).
correlations between variables was above 0.60, the variance of inflation (VIF) ranged from
indicate a low to moderate likelihood of issues related to multicollinearity affecting the
regression model (Field, 2009).
7.6.3 Mediation (Paths a and b), Indirect (ab), and Total Effects (c)
characteristics. The figures for these analyses can be found in Appendix H, along with the
tables of the Indirect and Direct Effects. Those that were significant are presented in the
following sections.
Overall, there were no fully or partially mediated relationships between the various
offender characteristics, motivations (mediators) and offence characteristics. Concerning the
indirect effects 44, the only relationship was found between sexual penetration behaviours and
44
The indirect effects of ab are the combined effects of paths a and b and their relationship with Y (Offender characteristics), independent of
the direct effects of X (offence characteristics) on Y (Hayes, 2009; Preacher & Hayes, 2008).
137
the offender having previous convictions, when the offence was motivated by Sexual
Compensation, as indicated by the confidence interval for this relationship, which did not
include zero, 0.001 to 0.204 (see Table 7.2). The enactment of the more deviant sexual
penetrative behaviours was associated with the Sexually Compensatory motivation, which in
turn was associated with the offender being more likely to have a history of previous
convictions.
Table 7.2
Indirect Effects of Sexual Penetration on Previous Convictions through Motivations (ab
paths)
Bootstrapping

CI
Angry 0.018 0.059 -0.072 0.158 -0.048 0.216 -0.054 0.203
Sadistic 0.016 0.161 -0.082 0.198 -0.049 0.514 -0.059 0.290
Sexually Compensatory 0.076 0.060 -0.008 0.185 0.001 0.200 0.003 0.204
Sexually Opportunistic -0.058 0.414 -1.588 0.041 -1.632 0.039 -1.461 0.056
Significant total effect relationships were found for perpetrator age, level of injury, and
offence outcome (Table 7.3); relationship status and offence outcome (just at significance
level, Table 7.4), and previous convictions, level of injuries, and victim age (Table 7.5). The
analysis indicates that an offence outcome of rape, as opposed to sexual murder, more likely
indicated an older offender, while sexual murder indicated a young offender. Lower levels of
victim injury was also associated with an older offender. It was also more likely that a rapist
would be in a sexual relationship at the time of their offence. A history of previous
convictions was tenuously associated with more severe injuries inflicted upon the victim by
the offender, as well as by, increased victim age. Again, these relationships are consistent
with those found in the previous chapters.
138
Table 7.3
Product of
Coefficients
B SE t p
Offence Outcome 5.479 1.963 2.791 0.007
Victim Age -0.040 0.054 -0.731 0.467
Weapon Used -2.771 2.444 -1.134 0.261
Table 7.4
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.888 0.457 1.941 0.052 3.769
Victim Age -0.014 0.012 -1.186 0.236 1.406
Weapon Used -0.417 0.529 -0.788 0.431 0.621
139
Table 7.5
Total Effects of Offence Variables on Previous Convictions (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome -0.051 0.535 -0.095 0.925 0.009
Victim Age 0.054 0.025 2.141 0.032 4.582
Weapon Used 0.005 0.633 0.007 0.994 0.000
7.6.4 Direct Effect of Offence Characteristics (X) on Offender Characteristics (Y) (c’)
The significant c’ path for perpetrator age was with regards to offence outcome. Perpetrator
age and offence outcome were positively associated, indicating that as the offender’s age
increased the offence was more likely to result in a rape, rather than a sexual murder. There
was one significant c’ path with regards to relationship status and offence outcome (also
found in Chapter 5 and 6) indicating that if the offence was a rape the offender was
significantly more likely to be in a relationship at the time of their offence. There were two
groups of significant c’ paths for the offender having any previous convictions: 1) level of
injury and 2) victim age. These findings, also found in Chapter 5 and 6, indicate that as the
level of injury increased, or as the victim age increased so did the likelihood that the offender
had previous convictions. There were no significant c’ paths for the offender living alone or
being employed at the time of their offence.
7.7 Discussion
The aim of the chapter was to test for mediating relationships between offender
characteristics, offence characteristics, and the offender’s motivations. There were no partial
and full mediating relationships found between these three groups of variables, suggesting
that the offender’s motivations and their offence characteristics are not associated in a
140
manner to be able to predict the offender’s characteristics from them. While, neither of the
motivations was found to have direct links between the offenders’ characteristics and offence
details, the four motivations were still identified within the current sample. Also, one indirect
relationship between the enactment of more deviant sexually penetrative behaviours and the
Sexually Compensatory motivation, and the offender’s history of previous convictions was
found. Despite the lack of causal relationships found in the current study, the influence of
motivations on the offending behaviour has been identified in previous studies (e.g., Beech et
al., 2005; Beech et al., 2006; Knight, 1999; Polaschek, Hudson, Ward, & Siegert, 2001;
Proulx & Beauregard, 2009a,b; Robertiello & Taylor, 2007; Vettor et al., in press).
As was found in Chapter 5 and 6, there were significant direct and total effect
relationships between offender age and offence outcome; relationship status and offence
outcome; and previous convictions and level of injury and victim age. Whilst, the inclusion of
the motivations did not have causal, or mediating, effects on the relationship between offence
and offender characteristics, they did affect the strength of, or moderate, the relationships
between these pairings (Baron & Kenny, 1986). The relationship between perpetrator age and
offence outcome was weakened by the inclusion of the motivations, as was the relationship
between relationship status and offence outcome. Whereas, the strength of the overall
associations between previous convictions, the level of victim injury, and the victim age were
improved when the presences of the motivations were considered.
The process of interfering characteristics from crime scene aspects relies on sound
theories that have been developed and tested empirically (Canter, 1995). The motivations
identified within the current sample have been repeatedly found and established in different
samples of rapists and sexual murderers (Beauregard & Proulx, 2002; Beech et al., 2006;
Beech et al., 2005; Brittian, 1970; Cohen et al., 1969; Gratzer & Bradford, 1995; Groth &
Birnbaum, 1979; Knight & Prentky, 1990; Knight et al., 1997; Proulx & Beauregard, 2009a;
Proulx et al., 1999; Warren, Hazelwood, & Dietz, 1996). These classifications have been
developed with the aim of distinguishing between different types of sex offenders based on
offender and crime characteristics, and limiting the heterogeneity of sexual offending into
more homogeneous classifications. The use of these homogeneous motivational
classifications can then be used to create a working profile of those offenders who sexual
assault which can be used to assist in the apprehension and sentencing of sexual offenders.
They can also be used in the development and implementation of treatment and the prediction
of future risk of re-offending as the motivations of sexual offender are very informative in
141
understanding the offender’s perceived reasons for their offending (Beech et al., 2006; Mann
& Hollins, 2007).
Outside of the motivational aspect, it was found that generally the older offender
committed rape, as opposed to sexual murder, inflicted less severe injuries on their victim and
was more likely to be in a sexual relationship. The infliction of more severe victim injury and
the presence of an older victim were both indicative of the offender having a history of
previous convictions. These findings combined with the motivational ones, start to provide an
outline of potential offender descriptions, which could be used as part of the greater
investigative process to help to elicit possible suspect lists, and potentially inform and target
police resources.
7.7.1 Limitations
A main drawback of the present study is the inability to be confident that the replies given to
the motivational aspects of the questionnaire and interview reflect the offender’s actual
motivations at the time of the offence. Their answers to the various questions may only
represent socially learned reasons for violating social laws, and their desire to hide some
aspects of their motivation in a bid to avoid being further ostracised for their transgressions
(Mann & Hollins, 2007). Despite this, the motivations were still evident in the current
sample, and were consistent with those found in other studies, which have used different data
collection methods, gives support to the validity of the existent and potential influence of the
identified motivations.
142
Chapter 8: General Discussion
8. Thesis Aims
The overall aim of the research presented in this thesis was to investigate the potential
mediating factors that may influence the ability to accurately predict offender characteristics
from crime scene characteristics; in other words, to look at potential influences on the A to C
process at the heart of Offender Profiling, and the ability to infer causal relationships between
offence and offender characteristics. The objective was to explore the influences of
contextual variables, perceptual variables, and motivational variables in an attempt to
understand the potential role they may play in how the offender commits their crime. The
intent was to provide a more holistic, accurate and pragmatically useful profile of an
unknown suspect which could in turn be used as part of an arsenal of investigative tools
utilised by law enforcement in their investigation of sexual offences.
8.1 Summary of Findings
A critical review and summary of the literature on the different approaches to Offender
Profiling found that each of the approaches, including the early and later theory-led
Pragmatic approach, the Clinical approach, and the Statistical approach, had their own unique
advantages, and underlying shortcomings. Regardless of the approach taken, there is mixed
support for either of the main assumptions underpinning offender profiling; more for
behavioural consistency than for homology. The lack of consistent support for these
assumptions, and offender profiling, may lie in the fact that the process of inferring offender
characteristics from offence is represented too simplistically, and does not address the
complexity of human behaviour (Ahadi & Diener, 1989; Mokros & Alison, 2002).
The rationale for the current thesis was to address the gaps in the research looking at
offender profiling and the ability to make inferences from crime scene variables to offender
characteristics, by identifying and testing potential factors that may influence offending
behaviour. The Cognitive-Affective Processing system (CAPS) developed by Mischel and
Shoda (1995; 1998) is used as a theoretical basis for the inclusion and exploration of various
potential mediating variables that may influence the ability to make appropriate inferences
about the offender from their crime scene. The inclusion of this model was in order to make
any findings comparable to previous literature as it has been used as theoretical support for
why the assumptions of, and offender profiling itself, could be viable in previous research
143
(e.g., Alison et al., 2002; Markson, Woodhams, & Bond, 2010; Sorochinski & Salfati, 2010).
The areas outlined for exploration in the current thesis were the context of the offence, such
as the location of contact with the victim, or the use of drugs or alcohol just prior to
offending, as well as the perceptions of the offender, which was investigated by identifying
the offender’s implicit theories. Lastly, the motivations of the offender is an area of research
which is often avoided in the Offender Profiling literature as they are hard to infer from crime
scene variables and their link back to offender characteristics is tenuous.
Often in the sex offending research, rapists and sexual murderers are investigated
separately, under the assumption there are significant differences between the two groups of
sexual offenders. Chapter 4 set out to explore the differences and/or similarities between
sexual murderers and rapists of adult women to determine if there was sufficient evidence of
similarities to combine the current sample of these offenders into one all encompassing
sample of sexual aggressors of adult women. Supporting previous research (e.g., Grubin,
1994; Milsom, Beech & Webster, 2003; Oliver, Beech, Fisher, & Beckett, 2007; Proulx,
Beauregard, Cusson, & Nicole, 2007), it was found that there were only a minimal number of
differences between the two categories of sexual offenders, indicating that rape and sexual
murder can be seen as a single model of sexual violence, and not as two separate models of
sexual homicide and rape (Salfati & Taylor, 2006). Since the two subgroups of sexual
aggressors differed on only a few variables they were combined for the statistical analyses
required to identify their pathways to offending. Past studies (Proulx & Beauregard, 2002,
2009; Proulx, Perreault, & Ouimet, 1999) have identified distinct pathways through multiple
correspondence analysis and clustering techniques using information about pre-crime and
modus operandi variables, victim and offender characteristics and situational variables with
Canadian samples of incarcerated sexual offenders. The pathways to offending of the current
combined sample of sexual aggressors were also examined, and three resulting pathways
were identified: the Angry, the Sadistic, and the Sexually Compensatory. These were similar
to those identified by Proulx and colleagues (Proulx, Perreault, & Ouimet, 1999; Proulx &
Beauregard, 2002; Proulx & Beauregard, 2009) with their Canadian samples, with the
addition of the Sexually Compensatory pathway and the omission of the Sexually
Opportunistic pathway due to its lack of replication in the current sample.
As suggested in Chapter 2, the three remaining empirical chapters (5,6, and 7),
examined the hypothesised mediators of the A to C equation of offender profiling. The
offender characteristics selected for inclusion in the analysis were perpetrator age,
relationship status, previous convictions, living and employment status, these represent the
144
type of variables predicted in offender profiles (Ault & Reese, 1980). The offence
characteristics selected for analysis were level of aggression, level of victim injury, offence
outcome, victim age, weapon use, and sexual penetration. Examining the relationship
between contextual variables, offender characteristics, and offence characteristics, Chapter
5explored the influence of the use of drugs or alcohol just prior to the offence, the location of
the initial contact with the victim, the location of the actual offence, and the level of victim
resistance. Some support was found for the influence of drugs just prior to the offence and the
location of the offence on the exhibition of various offence characteristics and aspects, such
as weapon use and victim age. These relationships were also linked with certain offender
characteristics, such as previous convictions and perpetrator age. Specifically, one partially
mediated relationship was found between victim age, location of offence, and the offender
having previous convictions, indicating that as the victim age increased, and if the offence
was indoors it was more likely that the offender had previous convictions. However, this was
only partially mediating, signifying that there was also a direct relationship between a
victim’s increasing age and the greater likelihood the offender had any previous convictions.
It was also found that there was a significant relationship between perpetrator age, drug use
and the use of a weapon during the offence, as well as, between perpetrator age, drug use, and
the enactment of more sexual deviant penetrative behaviours. Combined these tentatively
suggest that if the offender had used a weapon during their offence, or had enacted more
sexually deviant behaviours, they were more likely to be under the influence of drugs and
more likely to be older at the time of their offence. These findings were partially supported
by previous research.
Implicit Theories (ITs) (internal schemas which are used to explain, predict, and
interpret the behaviours, thoughts, and beliefs of others and the world) and how these may be
related to the offender’s characteristics and how they committed their offences was the focus
of Chapter 6. The same offence and offender characteristics used in Chapter 5 were used in
the analysis, with the five consistently found ITs, Dangerous World IT, Women as Sex
Objects IT, Entitlement IT, Women Unknowable/Dangerous IT, and Male Sex Drive is
Uncontrollable IT, also included. All but Women are Unknowable/Dangerous were indicated
in the current sample as the primary IT present, although none were found to play a
significant influence on the relationship between the offence and offender characteristics.
Even though, there were no significant relationships found between the variables, the ITs do
give insight into how beliefs and attitudes contribute to sexual offending as they provide
information about the cognition- behaviour relationship (Mihailides, Devilly, & Ward, 2004).
145
Although, it is unclear as to whether they initiate sexual offending or serve to maintain sexual
offending, and how they may interact with situational factors, so caution must be taken when
making conclusions about the role of ITs in the causation of offending behaviour.
Chapter 7 followed on from Chapters 5, and 6, in the exploration of different
mediating variables in the relationship between offender and offence characteristics,
focussing on the offender’s motivations for offending. The motivations explored were the
same as those focused on in Chapter 4 and previous research, namely angry, sadistic,
opportunistic, and compensatory. Despite the fact no significant relationships were found
between any of the motivations, offender characteristics, and offence characteristics,
motivations provide important information about the cognitive influences of sexual
offending, and can help to address the questions of what offenders were thinking when they
offended, and why they offended in the first place.
Direct relationships between crime scene aspects and offender characteristics were
found in all three chapters. Offence outcome, the offender’s age, and the offender’s
relationship status at time of offence were all related, with rapists being older (as opposed to
sexual murderers) and more likely to be in a sexual relationship at the time of their offence –
findings similar to those in Chapter 4 and supported by previous research (e.g., Grubin, 1994;
Milsom, Beech & Webster, 2003; Oliver, Beech, Fisher, & Beckett, 2007; Proulx,
Beauregard, Cusson, & Nicole, 2007). The offender having previous convictions was
positively associated with victim age, while the level of aggression indicated that the offender
was likely co-habiting at the time of their offence. Both of these relationships were supported
by previous research. Although, there were no significant findings, ITs and the offender’s
motivations do give insight into how beliefs and attitudes contribute to sexual offending, even
if it is unclear as to whether they initiate sexual offending or serve to maintain sexual
offending, and how they may interact with situational factors.
8.2 Theoretical and Practical Applications and Limitations
The findings in this thesis have important implications for Offender Profiling and potential
pragmatic usefulness in the investigation of sexual offences. Contextual aspects surrounding
offending, the perceptions of the offender, and their motivation behind offending were
individually explored to see which, if any, would play an influential role. The variables were
chosen based on their use in offender profiles and because of their potential predictive
validity as they have been found to affect the offending process.
146
8.2.1 Test of the Homology Assumption
As pointed out in Chapter 1 there is mixed support for the homology assumption of Offender
Profiling, very few consistent and direct relationships have been found between offence and
offender characteristics. The current thesis, did find support for the homology assumption.
Direct relationships were found across the chapters with regards to predicting offender age
from level of victim injury and offence outcome (rape/murder); relationship status and
offence outcome; previous convictions from level of victim injury, and victim age. The
relationship between younger offender age and the level of violent aggression, evident in
increasing victim injury and more likely outcome of murder, supports previous findings that
have shown how overtly aggressive offences are more likely committed by younger repeat
offenders (Gebhard, Gagnon, Pomeroy, & Christensen, 1965; Harry, Pierson, & Kuznetsov,
1993). The finding that rape indicated an offender who was more likely to be in a
relationship at the time of their offence, or the fact that sexual murderers were not, is
supported by the research showing more social isolation and difficulty of making and
maintaining close personal relationships for sexual murderers (Milsom et al., 2003).
8.2.1.1 Mediating Variables
There was only one partially mediated relationship found overall. This was between the
location of the offence, victim age, and previous offender convictions. The interaction
between these three variables is supported in previous literature (e.g., Feist et al., 2007;
Warren et al., 1998). What the majority of the findings seem to suggest, at least for the three
groups of mediators investigated in the current thesis, is that the inference of offender
characteristics from offence characteristics is not mediated by other variables, but potentially,
moderated by them. Whilst, there are direct one-to-one relationships between crime actions
and offender characteristics, the strength of these relationships are inhibited or facilitated by
situational, perceptual, and motivational factors. Therefore, the possibility of inferring
characteristics is not an all or nothing situation, but perhaps rests upon the abundance of
behavioural and contextual factors that shape predictive ability (Goodwin & Alison, 2007).
However, care must be observed when inferring causations for behaviour as the exact role
that contextual, cognitive, and motivational influences have on behavioural manifestations
and in offending behaviour still remains unclear.
147
8.2.1.2 Practical Applications
One of the main areas of pragmatic usefulness of Offender Profiling is as a behavioural

method to help narrow down the search for unknown offenders (Woodhams & Toye, 2007)
and the prioritisation of potential suspects possessing specified characteristics (Oldfield,
1997) which the current findings could provide limited support for. For example, in rape
investigations, the investigators could potentially prioritise older offenders, who are more
likely to in a sexual relationship. In addition, if the victim is older and the offence happened
indoors, this could indicate that the police look for a suspect who also has previous
convictions. If the offence is sexual murder, than it is more likely that the offender is
younger, and possibly not in a committed sexual relationship, with increased level of victim
injury also increasing the chance of a younger offender, and one who has previous
convictions.
The acknowledgment of situational factors may enhance the understanding of the
relationship between offender actions and characteristics (Mokros & Alison, 2002). If certain
situational and contextual factors are found to be associated with both crime scene behaviour
and offender characteristics this information could be used to help direct police resources at a
more targeted population of suspects. The current findings, at least tentatively, supports the
influence of offence location on the relationship between certain offence characteristics, such
victim age, and offender characteristics, such as having a previous criminal history. The
location of the offence may be linked to the offender’s regular daily activities, providing the
offender with potential opportunity for the identification of prospective victims (Cohen &
Felson, 1979), and providing the investigation with a location to start looking for information
about a possible suspect.
The inclusion of more cognitive aspects of the offence (e.g., perceptions, motivations) has
the potential to be beneficial for other areas of an investigation, such as suspect interview,
and understanding potential ‘change’ across the crime series of a single offender. The
inclusion of cognitive aspect can contribute to the understanding of the offender’s decision-
making process and their crime behaviour (Beauregard, Proulx, Rossmo, Leclerc, & Allaire,
2007). When deciding to commit a crime, an offender must decide (either implicitly or
explicitly) what the various options of action are as well as their corresponding consequences.
During this process, different situational and contextual variables can influence the
consequences of alternative courses of action and thus the decisions made and the actions
taken. These decisions are constrained by limits of time, opportunities, the offender’s
148
cognitive abilities, and the availability of relevant information (Cornish & Clarke, 1987;
Clarke & Felson, 1993). Additionally, the course of action decided upon is frequently the first
one that is gratifying or beneficial, and not necessarily the most optimal one (Proulx &
Beauregard, 2008). As such, the threat of punishment or the promise of a reward can motivate
an offender to commit a crime (Scott, 2000).All of which may serve to enhance the
reconstruction of the sexual offence (Beauregard & Fields, 2008).
The inclusion of contextual factors, routine activities, and the rational choices of
offenders attempts to explain offending through decisions of cost/benefit, the motivation and
interaction between person and behaviour, and how these are influenced by, and themselves
influence, the contextual and situational factors. The combination of which starts to build a
theoretical foundation underpinning of offender profiling, and a more complete
understanding of all aspects of the offender, their offending behaviour, and the interaction of
external constraints.
8.3 Limitations
Through each of the chapters the limitations regarding that study were explored, but an
overall limitation of the thesis is the often over simplification of the coded variables to
dichotomies. The main advantage of dichotomisation is the simplification of the results to an
easily understandable level, but the cost is a loss of information and richness of the data
(Farrington & Loeber, 2006), as well as the strength of association (Cohen, 1983) dependent
on the measure of strength used (Farrington & Loeber, 2006). Despite this, there are benefits
for criminological research because it does simplify results, making them understandable to a
wider audience, such as police, with only minimal effect on the resulting analysis, as long as
the right measures of association are used (e.g., tetrachoric) and the information lost is not
imperative to the research question (Farrington & Loeber, 2006). With regards to the current
thesis, there is a degree of relevant information lost through the dichotomisation of the
variables, although this type of coding is utilised quite often in offender profiling literature
due to limited nature of data access and collection techniques.
The present thesis was an exploratory study looking at the potential of mediating
factors in the ability to infer offender characteristics from offence characteristics. The
variables were chosen for their frequent inclusion in offender profiles and the information
contained in police files (Ault & Reese, 1980; Mokros & Alison, 2002; Ressler, Burgess, &
Douglas, 1988; Ressler & Douglas, 1985), but there remains some question as to how some
149
of the offence variables are supported by psychological theory in their ability to explain their
relationships with offender characteristics. Many of the studies testing the homology
assumption which consider potential influencing factors, such as situation and context, often
only find support, or provide support for, the offence characteristics to moderating variables
relationship (e.g., Beauregard, Proulx, Rossmo, Leclerc, & Allaire, 2007; Ullman, 2007;
Ward, Hudson, & Keenan, 1998). The CAPS model (Mischel & Shoda, 1995) helps to give
theoretical backing to the potential relationships between mediating or moderating variables
and offence characteristics, but the theoretical support for why there should be relationships
between mediators/moderators and offender characteristics, and between many offence
characteristics and many offender characteristics remains inadequate (Snook et al., 2008).
That said the link between potential mediators/moderators and offence characteristics is still
an important and useful one, which needs further examination.
8.4 Conclusions
In a top-down approach, the underlying assumption of homology was tested against

individual mediating influences to determine their reliability in predicting offender
characteristics. Despite the fact that the approach was not largely successful in finding
mediating relationships, there were indications of direct variable-to-variable relations, which
were potentially moderated by the proposed mediators. The importance of going beyond a
purely descriptive model of the relationship between offender traits, characteristics, and
behaviours, and looking at the potential role that situational, cognitive, affective, and
motivational variables play in the offending process is still very much needed. The role of
mediating and/or moderating influences is a new area of research surrounding Offender
Profiling. This thesis provides a glimpse into how the situation and context in which a sexual
crime happens, and how the offender perceives his world and those around him, and his
motivations for offending can influence the committal of the crime, and as well how these
influences can interact themselves with the individual’s characteristics to produce behaviour.
The future of offender profiling is an ever-changing prospect. More studies are being
conducted to evaluate the methodology of “profilers”, question long-standing ideals, employ
‘new’ statistical methods, and investigate the development and use of a multi-disciplinary
approach. A multi-disciplinary approach sees profilers and psychologists as part of the
investigative process working alongside the investigation. The knowledge they bring to any
150
investigation is a supplement to the already wide range of skills investigators already possess
(Jackson, van de Eshof, & de Kleuver, 1997).
151
References
Abbey, A. (1991). Acquaintance rape and alcohol consumption on college campuses: How
are they linked? American Journal of College Health, 39, 165-170.
Abel, G., & Osborn, C. (1992). The paraphilias: The extent and nature of sexually deviant
and criminal behaviour. Psychiatric Clinics of North America, 15, 675-687.
ACPO. (2006). Murder investigation manual. Wyboston: National Centre for Policing
Excellence.
Ahadi, S., & Diener, E. (1989). Multiple determinants and effect size. Journal of Personality
and Social Psychology, 56, 398-406.
Ainsworth, P. B. (2001). Offender profiling and crime analysis. Devon: Willan Publishing.
Alison, L., Bennell, C., Mokros, A., & Ormerod, D. (2002). The personality paradox in
offender profiling: A theoretical review of the processes involved in deriving
background characteristics from crime scene actions. Psychology, Public Policy and
Law, 8, 115-135.
Alison, L., & Canter, D. (1999). Profiling in policy and practice. Dartmouth: Aldershot.
Alison, L., Goodwill, A. M., & Alison, E. (2005). The madjenko, mascav and eve case: A
study in linking and suspect prioritisation. In L. J. Alison (Ed.), The forensic
psychologists casebook: A practical guide in preparing reports on violent and sexual
offences. London: Willian.
Alison, L., Goodwill, A., Almond, L., van den Heuvel, C., & Winter, J. (2010). Pragmatic
solutions to offender profiling and behavioural investigative advice. Legal and
Criminological Psychology,15, 115-132.
Alison, L., McLean, C., & Almond, L. (2007). Profiling suspects. In: T. Newburn, T.
Williamson, & A. Wright (Eds.). Handbook of Criminal Investigation. Devon: Willan.
Alison, L.J., Rockett, W., Deprez, S., & Watts, S. (2000). Bandits, cowboys and robin’s men:
The facets of armed robbery.. In D.V. Canter & L.J. Alison (Eds.), Profiling property
crimes (pp. 75-106). Aldershot, England: Ashgate Publishing.
152
Alison, L., Smith, M. D., Eastman, O., & Rainbow, L. (2003). Toulimn's philosophy of
argument and its relevance to offender profiling. Psychology, Crime, and Law, 9, 173-
183.
Alison, L., Smith, M. D., & Morgan, K. (2003). Interpreting the accuracy of offender
profiles. Psychology, Crime, & Law, 9, 185-195.
Alison, L.J., & Stein, K.L. (2001). Vicious circles: Accounts of stranger sexual assault reflect
abusive variants of conventional interactions. The Journal of Forensic Psychiatry, 12,
515-538.
Alison, L., Snook, B., & Stein, K. (2001). Unobtrusive measurement: Using police
information for forensic research. Qualitative Research, 1, 241-254.
Alison, L., West, A., & Goodwill, A. (2004). The Academic and the Practitioner:
Pragmatists’ views of Offender Profiling. Psychology, Public Policy and Law, 10, 71-
101.
Allport, G.W. (1961). Pattern and growth in personality. New York: Holt.
Almond, L., Alison, L., & Porter, L. (2007). An evaluation and comparison of claims made in
behavioural investigative advice reports compiled by the national policing
improvements agency in the united kingdom. Journal of Investigative Psychology and
Offender Profiling, 4, 71-83.
Alwin, D.F., Cohen, R.L., & Newcomb, T.M. (1991). Political attitudes over the life span:
The Bennington women after fifty years. Madison, WI: University of Wisconsin Press.
American Psychiatric Association (APA). (2000). Diagnostic and statistical manual of

mental disorders fourth edition, text revision). Author: Washington, DC.
Amir, M. (1971). Patterns of forcible rape. Chicago: University of Chicago Press.
Arrigo, B.A., & Purcell, C.E. (2001). Explaining paraphilias and lust murder: Toward an
integrated model. International Journal of Offender Therapy and Comparative
Criminology, 45, 6-31.
Ault, R., & Reese, J. (1980). A psychological assessment of crime: Profiling. FBI Law
Enforcement Bulletin , 49, 22-45.
153
Barbaree, H.E., Seto, M.C., Serin, R.C., Amos, N.L., & Preston, D.L. (1994). Comparisons
between sexual and nonsexual rapist subtypes: Sexual arousal to rape, offense
precursors, and offense characteristics. Criminal Justice and Behavior, 21, 95-114.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal
of Personality and Social Psychology, 51, 1173-1182.
Bartol, C. (1996). Police psychology: Then, now and beyond. Criminal Justice and Behavior,
23, 70-89.
Beachy, G.M., Petersen, D.M., & Pearson, F.S. (1979). Adolescent drug use and delinquency:
A research note. Journal of Psychedelic Drugs, 11,313-316.
Beauregard, E. (2010). Rape and sexual assault in investigative psychology: The contribution
of sex offenders’ research. Journal of Investigative Psychology and Offender
Profiling, 7, 1-13.
Beauregard, E., & Field, J. (2008). Body disposal patterns of sexual murderers: Implications
for Offender Profiling. Journal of Police and Criminal Psychology, 23, 81-89.
Beauregard, E., & Leclerc, B. (2007). An application of the rational choice approach to the
offending process of sexual offenders: A closer look at the decision-making. Sexual
Abuse: A Journal of Research and Treatment, 19, 115-133.
Beauregard, E., Lussier, P., & Proulx, J. (2009). Criminal propensity and criminal
opportunity: An investigation of crime scene behaviors of sexual aggressors of
women. In R.N. Kocsis (ed), Criminal profiling: International theory, research and
practice. (pp. 89-113). Totowa, NJ: Humana Press Inc.
Beauregard, E., & Proulx, J. (2002). Profiles in the offending process of nonserial sexual
murderers. International Journal of Offender Therapy and Comparative Criminology,
46, 386-399.
Beauregard, E., Proulx, J., Rossmo, K., Leclerc, B., & Allaire, J.F. (2007). Script analysis of
the hunting process of serial sex offenders. Criminal Justice and Behavior, 34, 1069-
1084.
154
Beauregard, E., Stone, M.R., Proulx, J., & Michaud, P. (2008). Sexual murderers of children:
Developmental, precrime, crime, and postcrime factors. International Journal of
Offender Therapy and Comparative Criminology, 52, 253-269.
Beech, A. R., Fisher, D., & Ward, T. (2005). Sexual murderers' implicit theories. Journal of
Interpersonal Violence, 20, 1366-1389.
Beech, A.R., Oliver, C., Fisher, D., & Beckett, R.C. (2005). STEP 4: The sex offender
treatment programme in prison: Addressing the needs of rapists and sexual
murderers. Available electronically from
www.hmprisonservice.gov.uk/assets/documents/100013DBStep_4_SOTP_report_2005.pdf
Beech, A. R., & Ward, T. (2004). The integration of aetiology and risk in sexual offenders: A
theoretical framework. Aggression and Violent Behavior, 10, 31-63.
Beech, A.R., Ward, T., & Fisher, D. (2006). The identification of sexual and violent
motivations in men who assault women: Implication for treatment. Journal of
Interpersonal Violence, 21, 1635-1653.
Bennell, C., & Canter, D. (2002). Linking commercial burglaries by modus operandi: Tests
using regression and ROC analysis. Science & Justice, 42, 153-164.
Bennell, C., & Jones, N. (2005). Between a ROC and a hard place. International Journal of
Investigative Psychology and Offender Profiling, 2, 23-41.
Bennel, C., Jones, N., Taylor, P., & Snook, B. (2006). Validities and abilities in criminal
profiling: A critique of the studies conducted by Richard Kocsis and his colleagues.
International Journal of Offender Therapy and Comparative Criminology, 50, 344-
360.
Berner, W., Berger, P., & Hill, A. (2003). Sexual sadism. International Journal of Offender
Therapy and Comparative Criminology, 47, 383-395.
Blackburn, R. (1993). The psychology of criminal conduct: Theory, research and practice.
Chichester, UK: Wiley.
Blackburn, R., & Fawcett, D. (1996). Manual for the antisocial personality questionnaire
(APQ). Unpublished manuscript, University of Liverpool, UK.
155
Blackburn, R., & Fawcett, D. (1999). The antisocial personality questionnaire: An inventory
for assessing personality deviation in offender populations. European Journal of
Psychological Assessment, 15, 14-24.
Blau, T. H. (1994). Psychological services for law enforcement. New York: Wiley.
Block, J., Block, J.H., & Keyes, S. (1988). Longitudinally foretelling drug usage in
adolescence: Early childhood personality and environmental precursors. Child
Development, 59, 336-355.
Block, R., & Skogan, W C. (1986). Resistance and nonfatal outcomes in stranger-to-stranger
predatory crime. Violence and Victims, 4,241-253.
Blumenthal, S., Gudjonsson, G., & Burns, J. (1999). Cognitive distortions and blame
attribution in sex offenders against adults and children. Child Abuse & Neglect, 23,
129-143.
Bollen, K. A., & Stine, R. (1990). Direct and indirect effects: Classical and bootstrap
estimates of variability. Sociological Methodology, 20, 115-140.
Boon, J. (1997). The contribution of personality theories to psychological profiling. In J.L.

Jackson, & D.A. Bekerian (Eds). Offender profiling: Theory, research, and practice
(pp. 44-59). Hoboken, NJ: John Wiley & Sons Inc.
Bowerman, B.L. and R.T. O’Connell. 1990. Linear statistical models: An applied approach.
Second Edition. Belmont, CA: Duxbury.
Bradford, J.M., Boulet, J., & Pawlak, A. (1992). The paraphilias: A multiplicity of deviant
behaviours. Canadian Journal of Psychiatry, 37, 104-108.
Brantingham, P., & Brantingham, P. (1981). Environmental criminology. Beverly

Hills, CA: Sage Publications
Brantingham, P., & Brantingham, P. (1984). Patterns in crime. New York, NY: Macmillan
Brecklin, L. R., & Ullman, S. E. (2001). The role of offender alcohol use in rape attacks: An
analysis of National Crime Survey data. Journal of Interpersonal Violence, 16, 3–21.
Brittain, R.P. (1970). The sadistic murderer. Medicine, Science and the Law, 10, 198-207.
156
Britton, P. (1997). The jigsaw man. Reading, England: Bantam Press.
Busch-Armendariz, N. B., DiNitto, D., Bell, H. & Bohman, T. (2010). Sexual assault
perpetrators’ alcohol and drug use: The likelihood of concurrent violence and post-
sexual assault outcomes for women victims. Journal of Psychoactive Drugs 42, 393-
399.
Bushman, B.J., & Cooper, H.M. (1990). Effects of alcohol on human aggression: An
integrative research review. Psychological Bulletin, 107, 341-354.
Burgess, A.W., Hartman, C.R., Ressler, R.K, Douglas, J.E., & McCormack, A. (1986).
Sexual homicide: A motivational model. Journal of Interpersonal Violence, 1, 251-
272.
Canter, D. (1989). Offender profiling, The Psychologist, 2, 12-16.
Canter, D. (1994). Criminal shadows. London: Harper Collins.
Canter, D. (1995a). Psychology of offender profiling. In R. Bull & D. Carson (Eds.),

Handbook of psychology in legal contexts (pp. 343-355). Chichester, UK: John Wiley
and Sons.
Canter, D. (1995b). Criminal shadows. London: Harper Collins.
Canter, D. (1996). A multivariate model of sexual offence behaviour: Developments in

‘Offender Profiling’. In Psychology in Action (pp. 189-216). Hantshire, UK:
Dartmouth Publishing Company.
Canter, D. (2004). Offender profiling and investigative psychology. Journal of Investigative

Psychology and Offender Profiling, 1, 1-15.
Canter, D. (2011). Resolving the offender “profiling equations” and the emergence of an
investigative psychology. Current Directions in Psychological Sciences, 20, 293-320.
Canter, D., Alison, L., Alison, E., & Wentink, N. (2004). The organized/disorganized
typology of serial murder. Psychology, Public Policy and Law, 10, 293-320.
Canter, D.V., Bennell, C., Alison, L.J., & Reddy, S. (2003). Differentiating sex offences: A
behaviorally based thematic classification of stranger rapes. Behavioral Sciences and
the Law, 21, 157-174.
157
Canter, D.V., & Fritzon, K. (1998). Differentiating arsonists: A model of fire-setting actions
and characteristics. Legal and Criminological Psychology, 3, 73-96.
Canter, D.V., & Heritage, R. (1990). A multivariate model of sexual offence behavior:
Developments in ‘offender profiling’. The Journal of Forensic Psychiatry, 1, 185-
212.
Canter, D.V., & Ioannou, M. (2004). A multivariate model of stalking behaviours.

Behaviormetrika, 31, 113-130.
Canter, D.V., & Wentink, N. (2004). An empirical test of Holmes and Holmes’s serial
murder typology. Criminal Justice and Behavior, 31, 489-515.
Canter, D., Heritage, R., Wilson, M., Davies, A., Kirby, S., Holden, R., et al. (1991). A facet
approach to offender profiling: Vol. 1. Guilford, England: University of Surrey,
Psychology Department.
Canter, D., Hughes, D., & Kirby, S. (1998). Paedophilia: pathology, criminality, or both? The
development of a multivariate model of offence behaviour in child sexual abuse. The
Journal of Forensic Psychiatry, 9, 532-555.
Caspi, A. (2000). The child is father of the man: Personality continuities from childhood to
adulthood. Journal of Personalty & Social Psychology, 78, 158-172.
Caspi, A., & Roberts, B.W. (2001). Personality development across the life course: The
argument for change and continuity. Psychological Inquiry, 12, 49-66.
Caspi, A., & Silva, P.A. (1995). Temperamental qualities at age 3 predict personality traits in
young adulthood: Longitudinal evidence from a birth cohort. Child Development, 66,
486-498.
Cattell, R.B. (1943). The description of personality: Basic traits resolved into clusters.
Journal of Abnormal and Social Psychology, 38, 476-506.
Cattell, R.B. (1945a). The description of personality: Principles and findings in a factor
analysis. American Journal of Psychology, 58, 69-90.
Cattell, R.B. (1945b). The principal trait clusters for describing personality. Psychological
Bulletin, 42, 129-161.
158
Cervone, D. (2005). Personality architecture: Within-person structures and processes. Annual
Review of Psychology, 56, 432-452.
Chaiken, J.M., & Chaiken, M.R. (1982). Varieties of criminal behavior: Summary and policy
implications. Santa Monica, CA: Rand Corporation
Chaney, E.F., O’Leary, M.R., & Marlatt, G.A. (1978). Skill training with alcoholics. Journal
of Consulting and Clinical Psychology, 46, 1092-1104.
Ch ѐnѐ, S., & Cusson,

aggressors: IntentionM. (2007). Sexual murde
and
situation. In J. Proulx, E. Beauregard, M. Cusson, & A. Nicole (Eds), Sexual
murderers: A comparative analysis and new perspectives (pp. 71-86). Chichester,
UK: John Wiley & Sons Ltd.
Chermack, S.T., & Blow, F.C. (2002). Violence among individuals in substance abuse
treatment: the role of alcohol and cocaine consumption. Drug and Alcohol
Dependence, 66, 29–37.
Chermack, S.T., Grogan-Kaylor, A., Perron, B.E., Murray, R.L., De Chavez, P., & Walton,
M.A. (2010). Violence among men and women in substance use disorder treatment: A
multi-level event-based analysis. Drug and Alcohol Dependence, 112, 194-200
Chesire, J.D. (2004). Review, critique, and synthesis of personality theory in motivation to
sexually assault. Aggression and Violent Behaviour, 9, 633-644.
Clarke, J., & Carter, A. J. (2000). Relapse prevention with sexual murderers. In D. R. Laws,
S. M. Hudson & T. Ward (Eds.), Remaking relapse prevention with sex offenders (pp.
389-401). London: Sage.
Clarke, R.V., & Felson, M. (1993). Routine activity and rational choice. Advances in
criminological theory. New Brunswick, NJ: Transaction Books.
Clausen, S. (1998). Applied correspondence analysis: An introduction. Thousand Oaks, CA:

Sage.
Cohen, J. (1983). The costs of dichotomization. Applied Psychological Measurement, 7, 249–

253.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159.
159
Cohen, L. E., & Felson, M. (1979). Social change and crime rate trends: A routine activities
approach. American Sociological Review, 44, 588-608.
Cohen, L.J., Garfalo, R.F., Boucher, R., & Seghorn, T. (1971). The psychology of rapists.
Seminars in Psychiatry, 3, 307-327.
Cohen, M.L., Seghorn, T., & Calmas, W. (1969). Sociometric study of sex offenders. Journal
of Abnormal Psychology, 74, 249-255.
Coleman, C., & Norris, C. (2000). Introducing criminology. Devon: William Publishing.
Collins, J.J. (1981). Drinking and crime – Perspectives on the relationship between alcohol
consumption and criminal behavior. New York, NY: Guildford Publications, Inc.
Collins, J.J., & Schlenger, W.E. (1988). Acute and chronic effects of alcohol use on violence.
Journal of Studies on Alcohol, 49, 516-521.
Collins, P., Johnson, G., Choy, A., Davidson, K., & MacKay, R. (1998). Advances in violent
crime analysis and law enforcement: The Canadian Violent Crime Linkage Analysis
System. Journal of Government Information, 25, 277-284.
Copson, G. (1995). Coals to Newcastle? Police use of offender profiling. London: Home
Office Police Research Group.
Copson, G., Badcock, R., Boon, J., & Britton, P. (1997). Editorial: Articulating a systematic
approach to clinical crime profiling. Criminal Behaviour and Mental Health, 7, 13-17.
Cornish, D., & Clarke, R. (1987). Understanding crime displacement: An application of

rational choice theory. Criminology, 25, 933-948.
Costa, P.T., & McCrae, R.R. (1990). Personality disorders and the five-factor model of
personality. Journal of Personality Disorders, 4, 362-371.
Costa, P.T., & McCrae, R.R. (1992). NEO PI-R Professional manual. Odessa, FL:
Psychological Assessment Resources.
Coucelis, H., Golledge, R., Gale, N., & Tobler, W. (1987). Exploring the anchor point
hypothesis of spatial cognition. Journal of Environmental Psychology, 7, 99-122.
160
Crabbé, A., Decoene, S., & Vertommen, H. (2008). Profiling homicide offenders: A review
of assumptions and theories. Aggression and Violent Behavior, 13, 88-106.
Daffern, M., Ferguson, M., Ogloff, J., Thomson, L., & Howells, K. (2007). Appropriate
treatment targets or products of a demanding environment? The relationship between
aggression in a forensic psychiatric hospital with aggressive behaviour preceding
admission and violent recidivism. Psychology, Crime & Law, 13, 431-441.
Dancer, L.S. (1990). Introduction to facet theory and its applications. Applied Psychology: An
International Review, 39, 365-377.
Davies, A. (1991). The use of DNA profiling and behavioural science in the investigation of
sexual offences. Medicine, Science and Law, 31, 95-101.
Davies, A. (1992). Rapists’ behaviour: A three aspect model as a basis for analysis and the
identification of serial crime. Forensic Science International, 55, 173-194.
Davies, A. (1999). Criminal personality profiling and crime scene assessment: A

contemporary investigative tool to assist law enforcement public safety. Journal of
Contemporary Criminal Justice, 15, 291-301.
Davies, A., Wittebrood, K., & Jackson, A. (1997). Predicting the criminal antecedents of a
stranger rapist from his offence behaviour. Science & Justice, 37, 161-170.
Davies, A., Wittebrood, K., & Jackson, J. L. (1998). Predicting the criminal record of a
stranger rapist (Special interest series paper 12). London: Home Office, Policing and
Reducing Crime Unit.
Davis, W.M. (1996). Psychopharmacologic violence associated with cocaine abuse: Kindling
of a limbic dyscontrol syndrome? Progress in Neuro-psychopharmacology &
Biological Psychiatry, 20, 1273–1300.
Denno, D.W. (1986). Victim, offender, and situational characteristics of violent crime. The
Journal of Criminal Law & Criminology, 77, 1142-1158.
DeRaad, B., Perugini, M., Hrebickova, M., & Szarota, P. (1998). Lingua franca of
personality: Taxonomies and structures based on the psycholexical approach. Journal
of Cross-cultural Psychology, 29, 212-232.
161
Deu, N., & Edelmann, R.J. (1997). The role of criminal fantasy in predatory and opportunist
offending. Journal of Interpersonal Violence, 12, 18-29.
Dietz, P.E., Hazelwood, R.R., & Warren, J. (1990). The sexually sadistic criminal and his
offenses. The Bulletin of the American Academy of Psychiatry and the Law, 18, 163-
178.
Digman, J.M., & Takemoto-Chock, N.K. (1981). Factors in the natural language of
personality: Reanalysis and comparison of six major studies. Multivariate Behavioral
Research, 16, 149-170.
Doan, B., & Snook, B. (2008). A failure to find empirical support for the homology
assumption in criminal profiling. Journal of Police and Criminal Psychology, 23, 61-
70.
Douglas, J., Burgess, A., Burgess, A., & Ressler, R. (1992). Crime classification manual.
Lexington, MA: Lexington Books.
Douglas, J., Ressler, R., Burgess, A. W., & Hartman, C. (1986). Criminal profiling from
crime scene analysis. Behavioral Sciences & the Law, 4, 401-421.
Douglas, D.A., Harford, T.C., & Rosenstock, I.M. (1994). Alcohol, other drugs, and sexual
risk-taking among young adults. Journal of Substance Abuse, 6, 87-93.
Dowden, C., Bennell, C., & Bloomfield, S. (2007). Advances in offender profiling: A
systematic review of the profiling literature published over the past three decades.
Journal of Police and Criminal Psychology, 22, 44-56.
Eaton, N., South, S.C., & Krueger, R.F. (2009). The cognitive-affective processing system
(CAPS) approach to personality and the concept of personality disorder: Integrating
clinical and social-cognitive research. Journal of Research in Personality, 43, 208
-217.
Efron, B., & Tibshirani, R. (1986). Bootstrap methods for standard errors, confidence
intervals, and other measure of statistical accuracy. Statistical Science, 1, 54-77.
Egger, S. (1984). A working definition of serial murder and the reduction of linkage
blindness. Journal of Police Science and Administration, 12, 348-357.
162
Elliott, R.T. (2010). Examining the relationship between personality characteristics and
unethical behaviors resulting in economic crime. Ethical Human Psychology and
Psychiatry, 12, 269-276.
Everitt, B., Landau, S., & Leese, M. (2001). Cluster analysis (fourth edition). London:
Hodder Arnold.
Fargo, J.D. (2007). An empirical taxonomy of incarcerated male sexual offenders using finite
mixture modelling: Adult victims. In G. Bourgon, R.K. Hanson, J.D. Pozzulo, K.E.
Morton Bourgon, C.L. Tanasichuk (eds.) (170-174). Proceedings of the 2007 North
American Correctional & Criminal Justice Psychology Conference. Corrections
research: User report. Ottawa: Public Safety Canada.
Farrington, D.P., & Loeber, R. (2000). Some benefits of dichotomization in psychiatric and
criminological research. Criminal Behaviour and Mental Health, 10, 100-122.
FBI. (2008, 30 September 2008). Investigative Programs Critical Incident Response Group.
Retrieved 23 October, 2008, from http://www.fbi.gov/hq/isd/cirg/ncavc.htm
Felson, M. (1987). ‘Routine activities and crime prevention in the developing metropolis.
Felson, R., & Krohn, M.(1990). Motives for rape. Journal of Research in Crime and
Delinquency, 27, 222–242.
Felson, R.B., & Steadman, H.J. (1983). Situational factors in disputes leading to criminal
violence. Criminology, 21, 59-74.
Fendrich, M., Mackesy-Amiti, M.E., Goldstein, P., Spunt, B., & Brownstein, H. (1995).
Substance involvement among juvenile murderers: Comparisons with older offenders
based on interviews with prison inmates. The International Journal of the Addictions,
30, 1363-1382.
Ferguson, C.J., White, D.E., Cherry, S., Lorenz, M., & Bhimani, Z. (2003). Defining and
classifying serial murder in the context of perpetrator motivation. Journal of Criminal
Justice, 31, 287-292.
Field, A. (2009). Discovering statistics using SPSS (third edition). Sussex: Sage Publications
Ltd.
163
Fisher, D., & Beech, A.R. (2007). Identification of motivations for sexual murder. In J.
Proulx, E. Beauregard, M. Cusson, & A. Nicole (Eds.), Sexual murderers: A
comparative analysis and new perspectives (pp. 175-190). Chichester, UK: John
Wiley & Sons Ltd.
Fiske, S.T., & Taylor, S. (1991). Social cognition. New York: McGraw-Hill.
Forer, B.R. (1949). The fallacy of personal validation: A classroom demonstration of

gullibility. The Journal of Abnormal and Social Psychology. 44, 118-123.
Francis, B., & Soothill, K. (2000). Does sex offending lead to homicide? The Journal of
Forensic Psychiatry, 11, 49-61.
Frank, R. Andresen, M.A., & Brantingham, P. (2011). Criminal directionality and the
structure of urban form. Journal of Environmental Psychology, 32, 37-42.
Frank, R., Dabbaghian, V., Reid, A., Singh, S., Cinnamon, J., & Brantingham, P. (2011).
Power of criminal attractors: Modelling the pull of activity nodes. Journal of Arificial
Societies and Social Simulation, 14 (1), 6. Available electronically from
http://jasss.soc.surrey.ac.uk/14/1/6.html
Fritzon, K., Canter, D., & Wilton, Z. (2001). The application of an action system model to
destructive behaviour: The examples of arson and terrorism. Behavioral Sciences &
the Law, 19, 657-690.
Fritzon, K., & Ridgway, J. (2001). Near-death experience: The role of victim reaction in
attempted homicide. Journal of Interpersonal Violence, 16, 679-696.
Funder, D.C., & Colvin, C.R. (1991) Explorations in behavioural consistency: Properties of
person, situations, and behaviours. Journal of Personality and Social Psychology, 60,
773-794.
Furr, R.M., & Funder, D.C. (2004). Situational similarity and behavioural consistency:
Subjective, objective, variable-centered, and person-centered approaches. Journal of
Research in Personality, 38, 421-447.
Gebhard, P H., Gagnon, J.H., Pomeroy, W.B., & Christensen, C.V. (1965). Sex offenders: An
analysis of types. New York: Harper and Row.
164
Gee, D., & Belofastov, A. (2007). Profiling Sexual Fantasy: Fantasy in Sexual Offending and
the Implications for Criminal Profiling. In R. N. Kocsis (Ed.), Criminal profiling:
International theory, research and practice (pp. 49-71). Totowa, NJ: Humana Press
Inc.
Gendreau, P., Little, T., & Gobbin, C. (1996). A meta-analysis of the predictors of adult
offender recidivism: What works! Criminology, 34, 575-607.
Glenn, N.D. (1980). Values, attitudes, and beliefs. In O.G. Brim, Jr. & J. Kagan (eds).
Constancy and change in human development (pp. 596-640). Cambridge, MA:
Harvard University Press.
Goldstein, P.J. (1985). The drugs/violence nexus: A tripartite conceptual framework. Journal
of Drug Issues, 15, 493-506.
Goodwill, A. M., & Alison, L. (2007). When is profiling possible? Offence planning and
aggression as moderators in predicting offender age from victim age in stranger rape.
Behavioral Sciences & the Law, 25, 823-840.
Goodwill, A.M., Alison, L., & Humann, M. (2009). Multidimensional scaling and the
analysis of sexual offence behaviour: A reply to Sturidsson et al. Psychology, Crime
& Law, 15, 517-524.
Gratzer, T., & Bradford, J.M.W. (1995). Offender and offense characteristics of sexual
sadists: A comparative study. Journal of Forensic Sciences, 40, 450-455.
Green, E.J., Booth, C.E., & Biderman, M.D. (1976). Cluster analysis of burglary M/O’s.
Journal of Police Science and Administration, 4, 382-388.
Greenfeld, L. (1997). Sex offences and offenders: An analysis of data on rape and sexual
assault. Available electronically at www.crimemagazine.com/sex_crimes.htm.
Groth, A.N. (1977). The adolescent sexual offender and his prey. International Journal of
Offender Therapy and Comparative Criminology, 21, 249-254.
Groth, A. N., & Birnbaum, H. J. (1979). Men who rape: The psychology of the offender. New
York: Plenum Press.
165
Groth, A. N., Burgess, A. W., & Holmstrom, L. L. (1977). Rape: Power, anger, and sexuality.
The American Journal of Psychiatry, 134, 1239-1243.
Groth, N.A., & Hobson, W.F. (1997). The dynamics of sexual assaults. In L.B. Schlesinger
& E. Revitch (Eds.), Sexual dynamics of anti-social behavior. (second edition, pp.
158-170). Springfield, IL: Charles C. Thomas.
Grubin, D. (1994). Sexual murder. British Journal of Psychiatry, 165, 624-629.
Grubin, D., Kelly, P., & Brunsdon, C. (2001). Linking serious sexual assault through
behavior. London: Home Office, Research Development and Statistics Directorate.
Gudjonsson, G.H., & Copson, G. (1997). The role of the expert in criminal investigation. In
J.L. Jackson & D.A. Bekerain (Eds), Offender profiling: theory, research and
practice (pp. 61-76). West Sussex, England: John Wiley & Sons.
Guttmann, L. (1968). A general nonmetric technique for finding the smallest coordinate
space for a configuration of points. Psychometrika, 33, 469-506.
Guttman, R. & Greenbaum, C.W. (1998). Facet theory: Its development and current status.
European Psychologist, 3, 13-36.
Hackshaw, A. (2008). Small studies: Strengths and limitations. European Respiratory

Journal, 32, 1141-1143.
Hanson, R. K., & Bussiere, M.T. (1998). Predicting relapse: A metaanalysis of sexual
offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-
362.
Hanson, R.K., & Morton-Bourgon, K. (2005). The characteristics of persistent sexual

offenders: A meta-analysis of recidivism studies. Journal of Consulting and Clinical
Psychology, 73, 1154-1163.
Harry, B., Pierson, T.R, & Kuznetsov, A. (1993). Correlates of sex offender and offense traits
by victim age. Journal of Forensic Sciences, 38, 1068-1074.
Haukoos, J.S., & Lewis, R.J. (2005). Advanced statistics: Bootstrapping confidence intervals
for statistics with “difficult” distributions. Academic Emergency Medicine, 12, 360-
365.
166
Hayes, A.F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new
millennium. Communication Monographs, 76, 408-420.
Hazelwood, R.R., & Burgess, A.W. (1987). An introduction to the serial rapist research by
the FBI. FBI Law Enforcement Bulletin, 56, 16-24.
Hazelwood, R.R., Reboussin, R., & Warren, J.I. (1989). Serial rape: Correlates of increased
aggression and the relationship of offender pleasure to victim resistance. Journal of
Interpersonal Violence, 4, 65–78.
Hazelwood, R., & Warren, J. (2003). Linkage analysis: Modus operandi, ritual and signature
in serial sexual crime. Aggression and Violent Behavior, 8, 587-598.
Heth, C.D., Cornell, E.H., & Flood, T.L. (2002). Self-ratings of sense of direction and route
reversal performance. Applied Cognitive Psychology, 16, 309–324.
Hicks, S. J., & Sales, B. D. (2006). Criminal profiling: Developing an effective science and
practice. Washington, D.C.: American Psychological Association.
Hogan, R. (1991). Personality and personality measurement. In M. D. Dunnette & L. M.

Hough (Eds.), Handbook of industrial and organizational psychology (pp. 327–396).
Palo Alto, CA: Consulting Psychologists Press
Hogan, J., & Ones, D.S. (1997). Conscientiousness and integrity at work. In R. Hogan,
J.Johnson, & S. Briggs (Eds), Handbook of personality psychology (pp. 849-870). San
Diego, CA: Academic Press.
Holmes, R. M., & De Burger, J. (1988). Serial murder. London: Sage Publications.
Holmes, R.M., & Holmes, S.T. (1998). Serial murder (2nd ed.). Thousand Oaks, CA: Sage.
Homant, R. J., & Kennedy, D. B. (1998). Psychological aspects of crime scene profiling:
Validity research. Criminal Justice and Behavior, 25, 319-343.
Home Office (2004). Offender management caseload statistics 2003. England and Wales
December 2004 (Table 8.2). www.homeoffice.gov.uk/rds/hosb2004.html.
House, J. C. (1997). Towards a practical application of offender profiling: The RNC's

criminal suspect prioritization system. In J. L. Jackson & D. A. Bekerain (Eds.),
167
Offender profiling: Theory, research and practice (pp. 177-190). Chichester,
England: Wiley.
Howlett, J., Hanfland, K., & Ressler R. (1986). Violent Criminal Apprehension Program –
VICAP: A progress report. FBI Law Enforcement Bulletin, 55, 14-22.
Hyman, H.H. (1955). Survey design and analysis: Principles, case, and procedures. Glencoe,
IL: Free Press.
Jackson, J. L., van de Eshof, P., & de Kleuver, E. E. (1997). A research approach to offender
profiling. In J. L. Jackson & D. A. Bekerain (Eds.), Offender profiling: theory
practice and research (pp. 107-132). Chichester: Wiley & Sons Ltd.
Jackson, J. L., van Koppen, P. J., & Herbrink, J. C. M. (1993). Does the service meet the
needs? An evaluation of consumer satisfaction with specific profile analysis and
investigative advice as offered by the Scientific Research Advisory Unit of the
National Criminal Intelligence Division (CRI). The Netherlands: Netherlands Institute
for the Study of Criminality and Law Enforcement (NISCALE).
Jankowski, D. (2002). A beginner’s guide to the MCMI-III. Washington, DC: American

Psychological Association.
Jaworska, N., & Chupetlovska-Anastasova, A. (2009). A review of multidimensional scaling

(MDS) and its utility in various psychological domains. Tutorials in Quantitative
Methods for Psychology, 5, 1-10.
John, O.P., & Srivastava, S. (1999). The big five trait taxonomy: History, measurement, and
theoretical perspectives. In. L.A. Pervin, & O.P. John (Eds.), Handbook of
personality: Theory, and research (2nd edition) (pp.102-138). New York: Guildford
Press.
Jones, L. (2004). Offence paralleling behaviour (OPB) as a framework for assessment and
interventions with offenders. In A. Needs & G. Towl (Eds.), Applying psychology to
forensic practice (pp. 34-63). BPS Blackwell: British Psychological Society
Jones, L. (2010). History of the offence paralleling behaviour construct and related concepts.
In M. Daffern, L. Jones & J. Shine (Eds.), Offence paralleling behaviour: A case
168
formulation approach to offender assessment and intervention (pp.3 -24). Chichester,
UK: Wiley & Sons Ltd.
Kafka, M.P. (1997). Hypersexual desire in males: An operational definition and clinical
implications for males with paraphilias and paraphilia-related disorders. Archives of
Sexual Behavior, 26, 505-526.
Kenny, D. T., Keogh, T., & Seidler, K. (2001). Predictors of recidivism in Australian juvenile
sex offenders: Implications for treatment. Sexual Abuse: A Journal of Research and
Treatment, 13, 131-148.
Keppel, R. D., & Walter, R. (1999). Profiling killers: A revised classification model for
understanding sexual murder. International Journal of Offender Therapy and
Comparative Criminology, 43, 417-437.
Kleck, G., & Sayles, S. (1990). Rape and resistance. Social Problems, 37, 149-162.
Kilpatrick, D.G., Best, C.L., Veronen, L.J., Amick, A.E., Villeponteaux, L.A., & Ruff, G.A.
(1985). Mental health correlates of criminal victimization: A random community
survey. Journal of Consulting and Clinical Psychology, 53, 866-873.
Knight, R.A. (1999). Validation of a typology for rapists. Journal of Interpersonal Violence,
14, 303-330.
Knight, R.A., & Prentky, R.A. (1987). The developmental antecedents and adult adaptations
of rapist subtypes. Criminal Justice and Behavior, 14, 403-426.
Knight, R. A., Warren, J. I., Reboussin, R., & Soley, B. (1998). Predicting rapist type from
crime-scene variables. Criminal Justice and Behavior, 25, 46-80.
Kocsis, R.N. (2003). Criminal psychological profiling: Validities and abilities. International
Journal of Offender Therapy and Comparative Criminology, 47, 126-144.
Kocsis, R. N. (2004). Psychological profiling of serial arson offenses: An assessment of skills
and accuracy. Criminal Justice and Behavior, 31, 341-361.
Kocsis, R.N., Cooksey, R.W., & Irwin, H.J. (2002). Psychological profiling of offender
characteristics from crime behaviors in serial rape offences. International Journal
of Offender Therapy and Comparative Criminology, 46, 144-169.
169
Kocsis, R.N., & Irwin, H.J. (1997). An analysis of spatial patterns in serial rape, arson and
burglary: The utility of Circle Theory of environmental range to psychological
profiling. Psychiatry,Psychology and Law, 4, 195–206.
Kocsis, R. N., Irwin, H. J., & Hayes, A. F. (1998). Organised and disorganised criminal
behaviour syndromes in arsonists: A validation study of a psychological profiling
concept. Australian and New Zealand Journal of Psychiatry, Psychology, and Law, 5,
117-131.
Kocsis, R. N., Irwin, H. J., Hayes, A. F., & Nunn, R. (2000). Expertise in psychological
profiling: A comparative assessment. Journal of Interpersonal Violence, 15, 311-331.
Kocsis, R. N., & Palermo, G. B. (2007). Contemporary problems in criminal profiling. In R.

N. Kocsis (Ed.), Criminal profiling: International theory, research, and practice (pp.
327-345). Totowa, New Jersey: Humana Press.
Kraemer, G.W., Lord, W.D., & Heilburn, K. (2004). Comparing single and serial homicide
offenses. Behavioral Sciences and the Law, 22, 325-343.
Kuznetsov, A., Pierson, T. R., & Harry, B. (1992). Victim age as a basis for profiling sex
offenders. Federal Probation, 56, 34–38.
Labuschange, G. (2006). The use of linkage analysis as evidence in the conviction of the
Newcastle serial murderer, South Africa. Journal of Investigative Psychology and
Landis, J.R., & Koch, G.G. (1977). The measurement of observer agreement for categorical
data. Biometrics, 33, 159-174
Langevin, R. (2003). A study of the psychosexual characteristics of sex killers: Can we

identify them before it is too late? International Journal of Offender Therapy and
Langevin, R., Ben-Aron, M., Wright, P., Marchese, P., & Handy, L. (1988). The sex killer.
Annals of Sex Research, 1, 263-301.
Langevin, R., Paitich, D., & Russon, A.E. (1985). Are rapists sexually anomalous,
aggressive, or both? In R. Langevin (Ed.), Erotic preference, gender identity, and
aggression in men: New research studies (pp. 17-38). Hillsdale, NJ: Erlbaum.
170
Langton, C.M., & Marshall, W.L. (2001). Cognition in rapists: Theoretical patterns by
typological breakdown. Aggression and Violent Behavior, 6, 499-518.
Lazarsfeld, P.G. (1955). Interpretation of statistical relations as a research operation. In P.F.

Lazarsfeld, & M. Rosenberg (eds), The Language of Social Research: A Reader in the
Methodology of Social Research (p. 115-125). Glencoe, IL: Free Press.
Lingoes, J. (1973). The Guttman-Lingoes nonmetric program series. Ann Arbor, MI:
Mathesis.
Lingoes, J. (1979). Geometric representation of relational data. Ann Arbor, MI: Mathesis
Press.
Luckenbill, D.F. (1981). Generating compliance: The case of robbery. Urban Life, 10, 25-46.
Ludwig, J.A., & Reynolds, J.F. (1988). Statistical ecology: a primer of methods and
computing. New York, NY: Wiley Press.
Lundrigan, S., & Canter, D. (2001). A multivariate analysis of serial murderers’ disposal site
location choice. Journal of Environmental Psychology, 21, 423-432.
MacCulloch, M.J., Snowden, P.R., Wood, P.J., & Mills, H.E. (1983). Sadistic fantasy,
sadistic behaviour and offending. The British Journal of Psychiatry, 143, 20-29.
MacKinnon, D.P., Fairchild, A.J., & Fritz, M.S. (2007). Mediation analysis. Annual Review
of Psychology, 58, 593-614.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A
comparison of methods to test mediation and other intervening variable effects.
Psychological Methods, 7, 83-104.
MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the
indirect effect: Distribution of the product and resampling methods. Multivariate
Behavioral Research, 39, 99-128.
Malamuth, N. (1986). Predictors of naturalistic sexual aggression. Journal of Personality and

Social Psychology, 50, 953-962.
171
Mann, R., & Beech, A.R. (2003). Cognitive distortions, schemas and implicit theories. In T.
Ward, D.R. Laws, & S.M. Hudson (Eds.), Theoretical issues and controversies in
sexual deviance (pp. 135-153). London: Sage.
Mann, R. E., & Hollin, C. R. (2007). Sexual offenders’ explanations for their offending.
Journal of Sexual Aggression, 13, 3-9.
Markson, L., Woodhams, J., & Bond, J.W. (2010). Linking serial residential burglary:
comparing the utility of modus operandi behaviours, geographical proximity, and
temporal proximity. Journal of Investigative Psychology and Offender Profiling, 7,
91-107.
Marlatt, G.A. (1982). Relapse prevention: A self-control program for the treatment of
addictive behaviours. In R.B. Stuart (Ed.), Adherence, compliance, and
generalization in behavioural medicine (pp. 329-378). New York: Brunner/Mazel.
Marlatt, G.A., & Gordon, J. (1980). Determinants of relapse: Implications for the
maintenance of change. In P.O. Davidson & S.M. Davidson (Eds.), Behavioral
medicine: Changing health lifestyles (pp. 410-452). New York: Brunner/Mazel.
Marlatt, G.A., & Gordon, J.R. (Eds.). (1985). Relapse prevention. New York: Guilford Press.
Marshall, W.L. (1989). Intimacy, loneliness and sexual offenders. Behavioral Research in
Therapy, 27, 491-503.
Marshall, W.L., & Kennedy, P. (2003). Sexual sadism in sexual offenders: An elusive
diagnosis. Aggression and Violent Behavior, 8, 1-22.
Matthews, H. (1995). Culture, environmental experience and environmental awareness:

Making sense of young Kenyan children’s views of place. The Geographical Journal,
3, 285–296.
McCabe, M.P., & Wauchope, M. (2005). Behavioural characteristics of rapists. Journal of

Sexual Aggression, 11, 235-247.
McCann, K., & Lussier, P. (2008). Antisociality, sexual deviance, and sexual reoffending in
juvenile sex offenders: A meta-analytical investigation. Youth Violence and Juvenile
Justice, 6, 363-385.
172
McDermott, M.J., & Hindelang, M.J. (1981). Juvenile criminal behavior: An analysis of rates
and victim characteristics. National Institute for Juvenile Justice and Delinquency
Prevention. Washington, D.C.: Government Printing Office.
Meaney, R. (2004). Commuters and marauders: An examination of the spatial behaviour of

serial criminals. Journal of Investigative Psychology and Offender Profiling, 1, 121-
137.
Meloy, J.R. (2000). The nature and dynamics of sexual homicide: An integrative review.
Aggression and Behavior, 5, 1-22.
Mendoza-Denton, R., Shoda, Y., Ayduk, O., & Mischel, W. (1999). Applying cognitive-
affective processing system (CAPS) theory to cultural differences in social behaviour.
In W.J. Lonner & D.L. Dinnel (Eds.). Merging past, present, and future in cross-
cultural psychology: Selected papers from the 14th international congress of the
International Association for Cross-Cultural Psychology (pp. 205-217). Lisse, The
Netherlands: Swets & Zeithinger.
Mendoza-Denton, R., & Mischel, W. (2007). Integrating system approaches to culture and
personality: The cultural cognitive-affective processing system.In S. Kitayama & D.
Cohen (Eds.), Handbook of Cultural Psychology (pp. 175-195). New York, NY: The
Guildford Press.
Mercado, R.G., & Paez, A. (2009). Determinants of distance travelled with a focus on the
elderly: A multilevel analysis in Hamilton CMA, Canada. Journal of Transport
Geography, 17, 65-76.
Mian, M., Wehrspann, W., Klajner-Diamond, H., LeBaron, D., & Winder, C. (1986). Review
of 125 children six years of age and under who were sexually abused. Child Abuse
and Neglect, 10, 223-229.
Mihailides, S., Devilly, G.J., & Ward, T. (2004). Implicit cognitive distortions and sexual
offending. Sexual Abuse: A Journal of Research and Treatment, 16, 333-350.
Milligan, G.W., & Cooper, M.C. (1987). Methodology review: Clustering methods.
Applied Psychological Measurement, 11, 329-354.
173
Millon, T. (1994). Millon Clinical Multiaxial Inventory-III manual. Minneapolis MN:
National Computer Systems.
Milsom, J., Beech, A.R, & Webster, S.D. (2003). Emotional loneliness in sexual murderers:
A qualitative analysis. Sexual Abuse: A Journal of Research and Treatment, 15, 285-
296.
Mischel, W. (1968). Personality and assessment. New York: Wiley.
Mischel, W., Mendoza-Denton, R., & Hong, Y. (2009). Towards an integrative CAPS
approach to racial/ethnic relations. Journal of Personality, 77, 1365-1380.
Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality:

Reconceptualising situations, dispositions, dynamics, and invariance in personality
structure. Psychological Review, 102, 246-268.
Mischel, W., & Shoda, Y. (1998). Reconciling processing dynamics and personality
dispositions. Annual Review of Psychology, 49, 229-258.
Mischel, W., & Shoda, Y. (2008). Toward a unified theory of personality: Integrating
dispositions and processing dynamics with the cognitive-affective processing system.
In O.P John, R.W. Robins, & L.A. Pervin (Eds). Handbook of personality: Theory
and research (pg 208-241). New York, NY: Guildford Press.
Mischel, W., Shoda, Y., & Mendoza-Denton, R. (2002). Situation-behavior profiles as a

locus of consistency in personality. Current Directions in Psychological Science, 11,
50-54.
Mokros, A., & Alison, L. J. (2002). Is offender profiling possible? Testing the predicted
homology of crime scene actions and background characteristics in a sample of
rapists. Legal and Criminological Psychology, 7, 25-43.
Mooney, C.Z., & Duval, R.D. (1993). Bootstrapping: A nonparametric approach to

statistical inference. Newbury Park, CA: Sage Publications, Inc.
Morency, C., Paez, A., Roorda, M.J., Mercado, R., & Farber, S. (2011). Distance travelled in
three Canadian cities: Spatial analysis from the perspective of vulnerable population
segments. Journal of Transport Geography, 19, 39-50.
174
Mrazek, P.J., Lynch, M.A., & Bentovim, A. (1983). Sexual abuse of children in the United
Kingdom. Child Abuse and Neglect, 7, 147-153.
Muller, D. A. (2000). Criminal profiling: Real science or just wishful thinking? Homicide
Studies, 4, 234-264.
Murphy, C.M., Winters, J., O’Farrell, T.J., Fals-Stewart, W., Murphy, M. (2005). Alcohol
consumption and intimate partner violence by alcoholic men: Comparing violent and
nonviolent conflicts. Psychology of Addictive Behaviors, 19, 35–42.
Myers, R.H. (1990). Classical and modern regression with applications (second edition).
Boston: Duxbury/Thompson Learning.
Myers, T. (1986). An analysis of context and alcohol consumption in a group of criminal

events. Alcohol and Alcoholism, 21, 389-395
Myers, W.C., Husted, D.S., Safarik, M.E., & O’Toole, M.E. (2006). The motivation behind
serial sexual homicide: Is it sex, power, and control, or anger? Journal of Forensic
Sciences, 51, 900-907.
Nicole, A., & Proulx, J. (2007). Sexual murderers and sexual aggressors: Developmental
paths and criminal history. In J. Proulx, E. Beauregard, M. Cusson, & A. Nicole
(Eds.), Sexual murderers: A comparative analysis and new perspectives (pp. 29-50).
Chichester, UK: John Wiley & Sons Ltd.
Norman, W.T. (1963). Toward an adequate taxonomy of personality attributes: Replicated

factor structure in peer nomination personality ratings. Journal of Abnormal and
Social Psychology, 66, 574-583.
Oldfield, D. (1997). What help do the police need with their enquiries? In J.L. Jackson &
D.A. Berekian (Eds.), Offender profiling: Theory, research and practice (pp. 93-106).
Chichester, England: Wiley.
Oliver, C.J., Beech, A.R., Fisher, D., & Beckett, R.C. (2007). A comparison of rapists and
sexual murderers on demographic and selected psychometric measures. International
Journal of Offender Therapy and Comparative Criminology, 51, 298-312.
Owen, R. (1843). Lectures on the comparative anatomy and physiology of the invertebrate
animals. London, England: Longman, Brown, Green, Longmans.
175
Owen, R. (2007). On the Nature of Limbs: A discourse, edited by R. Amundson (with a
preface by B.K. Hall and introductory essays by R. Amundson, K. Padian, M.P.
Windsor and J. Coggon). Chicago: University of Chicago Press.
Palermo, G.B. (2003). Editorial: A more comprehensive approach to sexual offenders.

International Journal of Offender Therapy and Comparative Criminology, 47, 3-5.
Palermo, G.B, & Kocsis, R.N. (2005). Offender profiling: An introduction to the
sociopsychological analysis of violent crime. Springfield, Ill: Charles C. Thomas
Publisher Ltd.
Palmer, C.T. (1988). Twelve reasons why rape is not sexually motivated: A sceptical
examination. The Journal of Sex Research, 25, 512-530.
Palmer, C.T., DiBari, D.N., & Wright, S.A. (1999). Is it sex yet? Theoretical and practical
implications of the debate over rapists’ motives. Jurimetrics, 39, 271-282.
Pardue, A., & Arrigo, B.A. (2008). Power, anger, and sadistic rapists. International Journal
of Offender Therapy and Comparative Criminology, 52, 378-400.
Pernanen, K. (1991). Alcohol in human violence. New York, NY: Guildford Press.
Personality. (n.d.). In Merriam-Webster’s online dictionary. Retrieved from

http://www.merriam-webster.com/dictionary/personality
Pervin, L. A. (2002). Current controversies and issues in personality (Third ed.): John Wiley
& Sons Inc.
Pinizzott, A. J. (1984). Forensic psychology: Criminal personality profiling. Journal of Police

Science & Administration, 12, 32-40.
Pinizzott, A. J., & Finkel, N. J. (1990). Criminal personality profiling: An outcome and
process study. Law and Human Behavior, 14, 215-233.
Pithers, W.D. (1990). Relapse prevention with sexual aggressors: A method for maintaining
therapeutic gain and enhancing external supervision. In. W.L. Marshall, D.R. Laws, &
H.E. Barbaree (Eds.), Handbook of sexual assault: Issues theories and treatment of
the offender (pp. 343-361). New York: Plenum.
176
Pithers, W.D., Marques, J.K., Gibat, C.C., & Marlatt, G.A. (1983). Relapse prevention with
sexual aggressive: A self-control model of treatment and maintenance of change. In
J.G. Greer & I.R. Stuart (Eds.), The sexual aggressor: Current perspectives on
treatment (pp. 214-234). New York: Van Nostrand Reinhold.
Polaschek, D.L.L., & Gannon, T.A. (2004). The implicit theories of rapists: What convicted
offenders tell us. Sexual Abuse: A Journal of Research and Treatment, 16, 299-314.
Polaschek, D., Hudson, S., Ward, T., & Siegert, R. (2001). Rapists’ offense processes: A
preliminary descriptive model. Journal of Interpersonal Violence, 16, 523-544.
Polaschek, D., & Ward, T. (2002). The implicit theories of potential rapists: What our
questionnaires tell us. Aggression and Violent Behavior, 7, 385-406.
Polaschek, D., Ward, T., & Hudson, S. (1997). Rape and rapists: Theory and treatment.
Clinical Psychology Review, 17, 117-144.
Pollock, N.L., & Hashmall, J.M. (1991). The excuses of child molesters. Behavioral Sciences
and the Law, 9, 53-59.
Porter, L., & Alison, L.J. (2004). Behavioural coherence in violent group activity: An
interpersonal model of sexually violent gang behaviour. Aggressive Behavior, 30,
449-468.
Porter, S., Woodworth, M., Earle, J., Drugge, J., & Boer, D. (2003). Characteristics of sexual
homicides committed by psychopathic and nonpsychopathic offenders. Law and
Human Behavior, 27, 459-470.
Preacher, K.J., & Hayes, A.F. (2008). Asymptotic and resampling strategies for assessing and
comparing indirect effects in multiple mediator models. Behavior Research Methods,
40, 879-891.
Prentky, R., & Burgess, A. (2000). Forensic management of sexual offenders. New York:
Kluwer Academic/Plenum Publishers.
Prentky, R.A., Burgess, A.W., & Carter, D.L. (1986). Victim responses by rapist type: An
empirical and clinical analysis. Journal of Interpersonal Violence, 1, 73-98.
177
Proulx, J., Aubut, J., Perron, L., & McKibben, A. (1994). Troubles de la personnalite et viol:
Implications theoriques et cliniques [Personality disorders and violence: Theoretical
and clinical implications]. Criminologie, 27, 33-53.
Proulx, J., & Beauregard, E. (2002). Profiles of the offending process of nonserial sexual
murderers. International Journal of Offender Therapy and Comparative Criminology,
46, 386-399.
Proulx, J., & Beauregard, E. (2008). Decision making during the offending process: An
assessment among subtypes of sexual aggressors of women. In T.Beech, L.A. Craig,
& K.D. Browne (Eds). Assessment and treatment of sex offenders: A handbook (pp.
181-197). Chichester, UK: Wiley.
Proulx, J., & Beauregard, E. (2009a). Decision making during the offending process: An
assessment among subtypes of sexual aggressors of women. In A.R. Beech, L.A.
Craig, & K.D. Browne (Eds.), Assessment and treatment of sex offenders: A
handbook (pp. 181-197). Chichester, UK: John Wiley & Sons Ltd.
Proulx, J., & Beauregard, E. (2009b). Pathways in the offending process of extrafamilial
sexual aggressors of women. Paper presented at the 28th Annual Association for the
Treatment of Sexual Abusers Research and Treatment Conference, Dallas, Texas,
USA. October 2009.
Proulx, J., Beauregard, E., Cusson, M., & Nicole, A. (2007) Sexual murderers: A
comparative analysis and new perspectives. Chichester, UK: Wiley.
Proulx, J., Beauregard, E., & Nicole, A. (2002). Developmental, personality and situational
factors in rapists and sexual murderers of women. Paper presented at the Association
for the Treatment of Sexual Abusers, Montreal, Quebec, Canada. October 2002.
Proulx, J., Cusson, M., & Beauregard, E. (2007). Sexual murder: Definitions, epidemiology
and theories. In J. Proulx, E. Beauregard, M. Cusson, & A. Nicole (Eds.), Sexual
murderers: A comparative analysis and new perspectives (pp. 9-28). Chichester, UK:
John Wiley & Sons Ltd.
Proulx, J., Perreault, C., & Ouimet, M. (1999). Pathways in the offending process of
extrafamilial sexual child molesters. Sexual Abuse: A Journal of Research and
Treatment, 11, 117-129.
178
Proulx, J., St-Yves, M., Guay, J.P., & Ouimet, M. (1999). Les aggresseurs sexuels de
femmes: Scenarios delictuels et troubles de la personnalite [Sexual aggressors of
women: Offence scenarios and personality disorders]. In J. Proulx, M. Cusson, & M.
Ouimet (Eds.), Les violences criminelles (pp. 157-185). Sainte-Foy, Quebec, Canada:
Les Presses de l’Universite Laval.
Pyle, G. (1974). The spatial dynamic of crime (Research Paper No. 159), Department of
Geography, University of Chicago, Chicago.
Quinn, G. & Keough, M. (2002). Experimental design and data analysis for biologists.
Cambridge, UK: Cambridge University Press
Quinsey, V.L., & Chaplin, T.C. (1982). Penile responses to nonsexual violence among
rapists. Criminal Justics and Behavior, 9, 372-381.
Quinsey, V.L., & Upfold, D. (1985). Rape completion and victim injury as a function of
female resistance strategy. Canadian Journal of Behavioural Science, 17, 40-50
Rainbow, L. (2007). The role of behavioural science in criminal investigations. Forensic

Update, 88, 44-48.
Rainbow, L., & Gregory, A. (2009). Behavioural investigative advice: A contemporary view.
The Journal of Homicide and Major Incident Investigation, 5, 71-82.
Raine, A., Reynolds, C., Venables, P.H., Mednick, S.A., & Farrington, D.P. (1998).
Fearlessness, stimulation-seekingm and large body size at age 3 years as early
predispositions to childhood aggression at age 11 years. Archives of General
Psychiatry, 55, 745-751.
Rajaratnam, S. M. W., Redman, J. R., & Lenne, M. G. (2000). Intoxication and criminal
behaviour. Psychiatry, Psychology and Law, 7, 59–69.
Reiser, M. (1982). Police psychology: Collected papers. Los Angeles: LEHI Publishing
Company
Royal Canadian Mounted Police. (n.d.). RCMP Violent Crime Linkage System (ViCLAS).
Retrieved from http://www.rcmp-grc.gc.ca/viclas-salvac/index-eng.htm
179
Resignato, A.J. (2000). Violent crime: A function of drug use or drug enforcement?. Applied
Economics, 32, 681–688.
Ressler, R., Burgess, A., & Douglas, J. (1988). Sexual homicide: Patterns and motives.
New York: The Free Press.
Ressler, R.K., & Douglas, J.E. (1985). Crime scene and profile characteristics of organized
and disorganized murderers.FBI Law Enforcement Bulletin, 58, 18-25.
Rice, M.E., Harris, G.T., & Quinsey, V.L. (1990). A follow-up of rapists assessed in a
maximum-security psychiatric facility. Journal of Interpersonal Violence, 5, 435-448.
Roberts, B.W., & Del Vecchio, W.F. (2000). The rank-order consistency of personality traits
from childhood to old age: A quantitative review of longitudinal studies.
Psychological Bulletin, 132, 1-25.
Robertiello, G., & Terry, K.J. (2007). Can we profile sex offenders? A review of sex
offenders typologies. Aggression and Violent Behaviour, 12, 508-518.
Roberts, B.W., & Caspi, A. (2001). Personality development and the person-situation debate:
It’s déjà vu all over again. Psychological Inquiry, 12, 104-109.
Romesburg, H.C. (2004). Cluster analysis for researchers. Morrisville, NC: Lulu Press.
Roslund, B., & Larson, C.A. (1979). Crimes of violence and alcohol abuse in Sweden.
Substance Use & Misuse, 14, 1103-1115.
Roth, N. (1952). Factors in the motivation of sexual offenders. The Journal of Criminal Law,
Criminology, and Police Science, 42(5), 631-635.
Salfati, G.C., & Bateman, A.L. (2005). Serial homicide: An investigation of behavioural
consistency. Journal of Investigative Psychology and Offender Profiling, 2, 121-144.
Salfati, G., & Taylor, P. (2006). Differentiating sexual violence: A comparison of sexual
homicide and rape. Psychology, Crime and Law, 12, 107-125.
Sampson, R. J. (2001). Sociology of delinquency. In N. J. Smelser & P. B. Baltes (Eds.),

International encyclopaedia of the social & behavioural sciences (pp. 3380-3384).
Elsevier Science Ltd.
180
Santtila, P., Fritzon, K., & Tamelander, A. L. (2005). Linking arson incidents on the basis of
crime scene behavior. Journal of Police and Criminal Psychology, 19, 1–16.
Santtila, P., Hakkanen, H., Canter, D., & Elfgren, T. (2003). Classifying homicide offenders
and predicting their characteristics from crime scene behaviour. Scandinavian Journal
of Psychology, 44, 107-118.
Santtila, P., Laukkanen, M., Zappala, A., & Bosco, D. (2008). Distance travelled and offence
characteristics in homicide, rape, and robbery against business. Legal and
Criminological Psychology, 13,345-356.
Santtila, P., Pakkanen, T., Zappala, A., Bosco, D., Valkama, M., & Mokros, A. (2008).
Behavioural crime linking in serial homicide. Psychology, Crime and Law, 14, 245-
265.
Sarangi, S., & Youngs, D. (2006). Spatial Patterns of Indian Serial Burglars with Relevance
to Geographical Profiling. Journal of Investigative Psychology and Offender
Profiling, 3, 105-115.
Saucier, G., & Goldberg, L.R. (1996). Evidence for the Big Five in analyses of familiar
English personality adjectives. European Journal of Personality, 10, 61-77.
Scott, J. (2000). Rational choice theory. In G. Browning, A. Halcli, & F. Webster,

Understanding contemporary society: Theories of the Present (pp. 126-138). London,
UK: Sage Publications.
Scully, D., & Marolla, J. (1985). Riding the bull at Gilley’s: Convicted rapists describe the
rewards of rape. Social Problems, 32, 251-263.
Shackleford, T.K. (2002). Are young women the special targets of rape-murder? Journal of
Aggressive Behavior, 28, 224-232.
Sheldon, K., & Howitt, D. (2008). Sexual fantasy in paedophile offenders: Can any model
explain satisfactorily new findings from a study of internet and contact sexual
offenders? Legal and Criminological Psychology, 13, 137-158.
Shoda, Y., Mischel, W., & Wright, J. C. (1994). Intraindividual stability in the organization
and patterning of behavior: Incorporating psychological situations into the idiographic
analysis of personality. Journal of Personality and Social Psychology, 67, 674–687.
181
Shye, S., Elizur, D., & Hoffman, M. (1994). Introduction to facet theory. London: Sage
Publications.
Siegel, J.M., Sorenson, S.B., Golding, J.M., Burnam, M.A., & Stein, J.A. (1989). Resistance
to sexual assault: Who resists and what happens? American Journal of Public Health,
79, 27-31.
Simon, L.M.J., Sales, B., Kaszniak, A., & Kahn, M. (1992). Characteristics of child
molesters: Implications for the fixated-regressed dichotomy. Journal of Interpersonal
Violence, 7, 211-225.
Silva, J.A., Leong, G.B., & Ferrari, M.M. (2004). A neuropsychiatric developmental model
of serial homicide behavior. Behavioral Sciences and the Law, 22, 787-799.
Snook, B., Cullen, R.M., Bennell, C., Taylor, P.J., & Gendreau, P. (2008). The criminal
profiling illusion: What’s behind the smoke and mirrors? Criminal Justice and
Behavior, 35, 1257-1276.
Snook, B., Eastwood, J., Gendreau, P., Goggin, C., & Cullen, R. M. (2007). Taking stock of
criminal profiling: A narrative review and meta-analysis. Criminal Justice and
Behavior, 34, 437-453.
Snook, B., Luther, K., House, J.C., Bennell, C., & Taylor, P. (2012). The Violent Crime
Linkage Analysis System: A test of interrater reliability. Criminal Justice and
Behavior, 39, 607-619.
Snyder, C.R., Jae Shenkel, R., & Lowery, C.R. (1977). Acceptance of personality
interpretations: The “Barnum Effect” and beyond. Journal of Consulting and Clinical
Psychology, 45,104-114.
Sobel, M.E. (1982). Asymptotic confidence internals for indirect effects in structural equation
models. In S. Leinhardt (Ed.), Sociological Methodology (pp. 290-312). Washington,
DC: American Sociological Association.
Sobel, M. E. (1986). Some new results on indirect effects and their standard errors in
covariance structure models. In N. Tuma (Ed.), Sociological Methodology (pp.
159-186). Washington, DC: American Sociological Association.
182
Sorochinski, M., & Salfati, G.C. (2010). The consistency of inconsistency in serial homicide:
Patterns of behavioural change across series. Journal of Investigative Psychology and
Stermac, L., & Hall, K. (1989). Escalation in sexual offending: Fact or fiction? Annals of Sex
Research, 2, 153-162.
Steyvers, M. (2002). Multidimensional scaling. In Encyclopedia of Cognitive Science.

London, UK: Nature Publishing Group.
Stone, C.A., & Sobel, M.E. (1990). The robustness of estimates of total indirect effects in
covariance structure models estimated by maximum likelihood. Psychometrika, 55,
337-52.
Sturidsson, K., Langstrom, N., Grann, M., Sjostedt, G., Asgard, U., & Aghede, E. M. (2006).
Using multidimensional scaling for the analysis of sexual offence behaviour: A
replication and some cautionary notes. Psychology, Crime, and Law, 12, 221-230.
Tabachnick, B.G., & Fidell, L.S. (1996). Using Multivariate Statistics (3rd ed.). New York:
Harper Collins College Publishers Heights, MA: Allyn and Bacon
Tabachnick, B.G., & Fidell, L.S. (2001). Using multivariate statistics (4th ed.). Needham
Tan, P.N., Steinback, M., & Kumar, V. (2005). Introduction to data mining. Minnesota, MN:
Addison-Wesley.
Tapert, S.F., Aarons, G.A., Sedlar, G.R., & Brown, S.A. (2001). Adolescent substance use
and sexual risk-taking behaviour. Journal of Adolescent Health, 28, 181-189.
Taylor, P.J., Donald, I.J., Jacques, K., & Conchie, S.M. (2010). Jaccard’s heel: Radex
models of criminal behaviour are rarely falsifiable when derived using Jaccard
coefficient. Legal and Criminological Psychology, 15, 1-18.
Ter Beek, M., Van Den Eshof, P., & Mali, B. (2010). Statistical modelling in the
investigation of stranger rape. Journal of Investigative Psychology and Offender
Profiling, 7, 31-47.
183
Tinklenberg, J.R., & Ochberg, F.M. (1981). Patterns of adolescent violence: A California
sample. In D.A. Hamber, M.B. Trudeau (Eds). Bio-behavioral Aspects of Aggression
(pp. 121-140). New York, NY: Allan R. Liss Inc.
Tonkin, M., Bond, J.W., & Woodhams, J. (2009). Fashion conscious burglars? Testing the
principles of offender profiling with footwear impressions recovered at domestic
burglaries. Psychology, Crime & Law, 15, 327-345.
Turco, R. (1990). Psychological profiling. International Journal of Offender Therapy and

Ullman, S.E. (1998). Does offender violence escalate when rape victims fight back? Journal
of Interpersonal Violence, 13, 179-192.
Ullman, S.E., Karabatos, G., & Koss, M.P. (1999a). Alcohol and sexual assault in a national
sample of college women. Journal of Interpersonal Violence, 14, 603-625.
Ullman, S. E., Karabatsos, G., & Koss, M. P. (1999b). Alcohol and sexual aggression in a
national sample of college men. Psychology of Women Quarterly, 23, 673–689.
Ullman, S.E., & Knight, R.A. (1993). A multivariate model for predicting rape and physical
injury outcomes during sexual assaults. Journal of consulting and Clinical
Psychology, 59, 724-731.
Ullman, S.E., & Siegel, J.M. (1993). Victim-offender relationship and sexual assault.
Violence and Victims, 8, 121-134.
Van Mechelen, I., & De Raad, B. (1999). Editorial: Personality and situations. European
Journal of Personality, 13, 333-336.
Walker, A. (1985). The Care Gap: How Can Local Authorities meet the Needs of the
Elderly? London: Local Government Information Unit.
Ward, T. (2000). Sexual offenders' cognitive distortions as implicit theories. Aggression and
Violent Behavior, 5, 491-507.
Ward, T., & Hudson, S.M. (1998). A model of the relapse process in sexual offenders.
Journal of Interpersonal Violence, 13, 700-725.
184
Ward, T., Hudson, S. M., Johnston, L., & Marshall, W. L. (1997). Cognitive distortions in
sex offenders: An integrative review. Clinical Psychology Review, 17, 479-507.
Ward, T., Hudson, S.M., & Keenan, T. (1998). A self-regulation model of the sexual offense
process. Sexual Abuse: A Journal of Research and Treatment, 10, 141-157.
Ward, T., & Keenan, T. (1999). Child molesters’ implicit theories. Journal of Interpersonal
Violence, 14, 821-838.
Ward, T., Polaschek, D. L. L., & Beech, T. (2006). Theories of sexual offending. Chichester:
John Wiley and Sons, Ltd.
Warren, J.I., Hazelwood, R.R., &Dietz, P.E. (1996). The sexually sadistic serial killer.
Journal of Forensic Sciences, 41, 970-974.
Warren, J., Reboussin, R., Hazelwood, R., Gibbs, N., Trumbetta, S., & Cummings, A. (1999).
Crime scene analysis and the escalation of violence in serial rape. Forensic Science
International, 100, 37-56.
Warren, J., Reboussin, R., Hazelwood, R.R., Cummings, A., Gibbs, N., & Trumbetta, S.
(1998). Crime scene and distance correlates of serial rape. Journal of Quantitative
Weiner, M.D., Sussman, S., Sun, P., & Dent, C. (2005). Explaining the link between violence
perpetration, victimization and drug use. Addictive Behaviors, 30, 1261-1266.
Wilson, A., & Alison, L. (2004). Questioning sequences in Canadian police interviews:
Constructing and confirming the course of events?. Psychology, Crime and Law, 10,
137-154.
Wilson, M., Daly, M., & Scheib, J. (1997). Femicide: An evolutionary psychological
perspective. In A.P. Gowaty (Ed.), Feminism and evolutionary biology (pp. 431-465).
New York, NY: Chapman and Hall.
Wolfgang, M.E., & Strohm, R.B. (1956). The relationship between alcohol and criminal
homicide. Quarterly Journal of Studies on Alcohol, 17, 411-425.
Woodhams, J. (2012). Offender profiling and crime linkage. In G. Davies, & A. Beech (eds),
Forensic psychology: Crime, justice, law, interventions (2nd edition) (pp.171-188).
Chichester, UK: John Wiley & Sons Ltd.
185
Woodhams, J., Bull, R., & Hollin, C. R. (2007). Case linkage: Identifying crimes committed
by the same offender. In R. N. Kocsis (Ed.), Criminal profiling: International theory,
research and practice (pp. 117-133). Totowa, NJ: Humana Press Inc.
Woodhams, J. & Grant, T. (2006). Developing a categorization system for rapists’ speech.
Psychology, Crime and Law, 12, 245-260.
Woodhams, J., Hollin, C.R., & Bull, R. (2007). The psychology of linking crimes: A review
of the evidence. Legal and Criminological Psychology, 12, 233-249.
Woodhams, J., & Toye, K. (2007). An empirical test of the assumptions of case linkage and
offender profiling with serial commercial robberies. Psychology, Public Policy, and
Law, 13, 59-85.
Youngs, D. (2004). Personality correlates of offence style. Journal of Investigative

Psychology and Offender Profiling, 1, 99-119.
Zolondek, S., Abel, G., Northey, W., & Jordan, A. (2001). The self-reported behaviors of
juvenile sexual offenders. Journal of Interpersonal Violence, 16, 73-85.
186
Appendix A: Chapter 3 Dendogram Example (Chapter 4 Full Cluster Analysis without MCMI data)
Cluster Membership of the 102 Sexual Aggressors by Individual Offenders
187
Appendix B: Functional Analysis
Question
188
189
190
191
192
193
194
195
196
197
Appendix C: Chapter 4 Comparisons Tables
Table B1
Offender, Lifestyle, Victim, Pre-crime and Modus Operandi characteristics
Totala Sexual Murderers Rapists X2 df Phi
(%) (%) (%)
Offender and General Lifestyle Characteristics:
Ethnicity White 77.5 92.1 68.8 7.45** 1 0.27
Other 22.5 2.9 19.6 8.34* 3 0.29
African Caribbean 14.7 7.9 18.8

Asian 4.9 0.0 7.8
Other 2.9 0.0 4.7
Where offender was living Lived alone 19.6 21.1 18.8 0.08 1 0.03
Other 80.4 78.9 81.3
Parent(s) 21.6 28.9 17.2 7.96 8 0.28

Lived Alone 19.6 21.1 18.8
Wife/girlfriend 18.6 21.0 17.2
Other/Not known 40.3 29.0 46.9
Employment status Unemployed 39.2 39.5 39.1 0.00 1 0.97
Employed/student 60.8 60.5 60.1
Major life event Yes 84.3 81.6 85.9 0.34 1 0.06

No 15.7 18.4 14.1
Inter-relational difficulties 10.8 21.1 4.7 6.64* 1 0.26

Relationship break up 47.1 52.6 43.8
House move 31.4 23.7 35.9
198
Relationship and Sexual Lifestyle Characteristics:
In Relationship at time of offence Yes 65.7 52.6 73.4 4.58* 1 0.21
No 34.3 47.4 26.6
Relationship length Less than a year 56.9 71.1 48.4 4.97* 1 0.22
More than a year 43.1 28.9 51.6
Gender of current partner Male 34.3 44.7 28.1 2.92 1 0.17

Female 65.7 55.3 71.9
Dissatisfaction with current relationship Yes 55.9 42.1 64.1 4.66* 1 0.21
No 44.1 57.9 35.9
Previous sexual partners Use of prostitutes 8.8 7.9 9.4 0.07 1 0.03
Male 71.6 63.2 76.6 2.11 1 0.14

Female 28.4 36.8 23.4
Sexual relationships A few long term relationships 61.8 47.4 70.3 5.32* 1 0.23
Many short term relationships 40.2 36.8 42.2 0.28 1 0.05
Few short term relationships 42.2 47.4 39.1 0.68 1 0.08
Friendship characteristics Difficulty making close friends 37.3 50.0 29.7 4.21* 1 0.20
Problem keeping friends 24.8 31.6 20.6 1.52 1 0.12
Childhood Victimisation:
Sexual Abuse as Child Yes 29.4 8.8 20.6 6.68* 2 0.26
No 17.6 2.9 14.7
Unknown 52.9 25.5 27.5
Perpetrator of Sexual Abuseb Parent 23.1 60.0 14.3 4.75 1 0.43

Other 76.9 40.0 85.7
199
Parent 23.1 60.0 14.3 6.04 4 0.48
Family/Friend/Acquaintance 65.3 20.0 76.2
Stranger 11.5 20.0 9.5
Form of Sexual Abusec Shown Pornography 17.6 33.3 14.3 6.68 3 0.63
Witnessed Sex 5.9 33.3 0.0
Assault/masturbation/oral sex 41.2 0.0 50.0
Buggary/intercourse 35.3 33.3 35.7
Physically Abuse as Child Yes 34.3 6.9 27.5 0.26 1 0.08

No 10.8 2.9 7.8
Unknown 52.9 27.5 27.5
Perpetrator of Physical Abused Biological Parent 31.4 42.9 28.6 0.71 2 0.14
Step Parent 2.9 0.0 3.6
Unknown 65.7 57.1 67.9
Form of Physical Abusee Smacked/hit 17.1 0.0 21.4 6.04 4 0.42

Hit object 28.6 42.9 25.0
Punched/kicked 17.1 28.6 14.3
Extreme acts 22.9 0.0 28.6
Not specified 14.3 28.6 10.7
Previous Offence History:

Sexual Yes 38.2 28.9 43.8 2.21 1 0.15
No 61.8 71.1 56.3
Violent Yes 42.2 39.5 43.8 0.18 1 0.04

No 57.8 60.5 56.3
Other Yes 65.7 63.2 67.2 0.17 1 0.04

No 34.3 36.8 32.8
200
Any Previous Offence Yes 76.5 76.3 76.6 0.00 1 0.00
No 23.5 23.7 23.4
Pre-Crime Factors
Pressure Relationship 45.1 57.9 37.5 4.01* 1 0.20
Family 33.3 52.6 21.9 10.15** 1 0.31
Work 30.4 31.6 29.7 0.04 1 0.02
Affect just prior to offence Happy/content 22.5 21.1 23.4 0.08 1 0.03
Upset 52.0 50.0 53.1 0.09 1 0.03
Anxious 69.6 73.7 67.2 0.48 1 0.07
Humiliated 32.4 44.7 25.0 4.24* 1 0.20
Angry 62.7 71.1 57.8 1.79 1 0.13
Substance use at time of offence Drugs 38.2 28.9 43.8 2.21 1 0.15
Cocaine 14.7 2.6 21.9 7.04** 1 0.26

Cannabis 35.3 26.3 40.6 2.14 1 0.15
Amphetamines 20.6 15.8 23.4 0.85 1 0.09
Felt in control of life just prior to offence Yes 27.5 23.7 29.7 0.43 1 0.07
No 72.5 76.3 70.3
Fantasised about offence within 48hrs of offence Yes 51.0 34.2 60.9 6.82** 1 0.26
No 49.0 65.8 39.1
Victim Characteristics:
Relationship to offender Stranger 55.9 47.4 60.9 1.78 1 0.13
Known 44.1 52.6 39.1
Stranger 50.0 47.4 51.6 8.66* 3 0.29
201
Prostitutes 5.9 0.0 9.4
Current/past partner 15.7 10.5 18.8
Acquaintance/friend 28.4 42.1 20.3
Victim behaviour during offence Physically hostile 17.6 15.8 18.8 0.14 1 0.04
Verbally hostile 15.7 10.5 18.8 1.22 1 0.11
Talking offender out of offence 36.3 15.8 48.4 10.99** 1 0.33
Crying/pleading 30.4 28.9 31.3 0.06 1 0.02
Being compliant 51 31.6 62.5 9.12** 1 0.30
Modus Operandi:
Planned offence Yes 44.1 28.9 53.1 5.65* 1 0.24
No 55.9 71.1 46.9
Planned who to offend against Yes 16.7 5.3 23.4 5.67* 1 0.24
No 83.3 94.7 76.6
Plan where offence would happen Yes 14.7 7.9 18.8 2.24 1 0.15
No 85.3 92.1 81.3
Plan when offence would happen Yes 14.7 2.6 21.9 7.04** 1 0.26
No 85.3 97.4 78.1
Preparation Weapon brought with offender 24.5 28.9 21.9 0.65 1 0.08
Disguise 4.9 5.3 4.7 0.02 1 0.01
Gloves 9.8 15.8 6.3 2.45 1 0.16
Condom 3.9 0.0 6.3 2.47 1 0.16
Coercion used to overcome victim resistance Verbal threats 37.3 13.2 86.8 15.04*** 1 0.38
Threat with weapon 24.5 13.2 31.3 4.22* 1 0.20
Force, no serious injury 36.6 5.3 55.6 25.83*** 1 0.51
202
Force, serious injury 35.3 60.5 20.3 16.88*** 1 -0.41
Weapon Brought with 31.4 28.9 32.8 0.17 1 0.04

Found at scene 9.8 23.7 1.6 13.20** 1 0.36
Victim Incapacitated Restrained 39.2 52.6 31.3 4.57* 1 0.21

Unconscious 9.8 23.7 1.6 13.20*** 1 0.36
Physically held 15.7 13.2 17.2 0.29 1 0.05
Touch/penetrate vagina Penis 63.7 34.2 81.3 22.82*** 1 0.47

Finger 35.3 31.6 37.5 0.37 1 0.06
Foreign object 7.8 10.5 6.3 0.60 1 0.08
Touch/penetrate anus Penis 11.8 7.9 14.1 0.87 1 0.09
Inflict Pain Yes 30.4 36.8 26.6 1.19 1 0.11

No 69.6 63.2 73.4
Strangulation of victim Yes 4.9 10.5 1.6 4.11 1 0.20

No 95.1 89.5 98.4
*<0.05; **<0.01; ***<0.001
a. Unless otherwise noted N=102
b. N=26
c. N=17
d. N=34
e. N=35
Note: After correcting for Type I error, only p values < 0.001 are significant. These are highlighted in bold.
203
Table B2
Age Comparisons between Rapists and Sexual Murderers
Sexual
Total Murderers Rapists t df F p
M (SD) M (SD) M (SD)
Age at First Offence a 7-57 24.11(8.29) 22.03 (7.38) 25.34 (8.62) 1.84 87 3.85 0.05
Age at Offence b 14-57 26.70 (8.60) 23.88 (7.08) 29.36 (9.14) 2.79 68 3.25 0.08
Mdn (Range) Mdn (Range) Mdn (Range) U z r p

Age of Victim c 27.00 (16-86) 35.00 25.00 864.50 -2.44 -0.24 0.02
Age Sexual Abuse

12.00 (4-15) 12.00 (4-15) 12.00 (5-15) 46.50 -0.07 -0.01 0.96
started d
a. n=89
b. n=70
c. n=102
d. n=24
204
Appendix D: Figures and Tables of the Context Mediation Analyses
Perpetrator Age from Level of Aggression Used
-0.024 M 7.796
X Y
Model Summary: R2=0.026, F=0.900 p=0.441
-0.441
Figure A1. Predicting Perpetrator age (Y) from Level of Aggression (X) through Use of
Drugs (M).
0.082 -0.245
M
X Y
-0.606
Figure A2. Predicting Perpetrator age (Y) from Level of Aggression (X) through Use of
Alcohol (M).
0.096 -2.825
M
X Y
-0.355
Figure A3. Predicting Perpetrator age (Y) from Level of Aggression (X) through Location of
Initial Contact (M).
-11.810 M 0.002
X Y
-0.602
Figure A4. Predicting Perpetrator age (Y) from Level of Aggression (X) through Location of
Offence (M).
0.124 M -0.346
X Y
-0.583
Figure A5. Predicting Perpetrator age (Y) from Level of Aggression (X) through Victim
Resistance (M).
205
Perpetrator Age from Level of Injury
0.000 M 7.753
X Y
0.002
Figure A6. Predicting Perpetrator age (Y) from Level of Injury (X) through Use of Drugs
(M).
0.053 3.098
M
X Y
-2.227*
* p < 0.05
Figure A7. Predicting Perpetrator age (Y) from Level of Injury (X) through Use of Alcohol
(M).
0.004 -4.384
M
X Y
-2.047*
* p < 0.05
Figure A8. Predicting Perpetrator age (Y) from Level of Injury (X) through Location of
-22.511 0.001
M
X Y
-2.052*
* p < 0.05
Figure A9. Predicting Perpetrator age (Y) from Level of Injury (X) through Location of
Offence (M).
206
-0.087 0.262
M
X Y
-2.041*
*p=0.05
Figure A10. Predicting Perpetrator age (Y) from Level of Injury (X) through Victim
Resistance (M).
Perpetrator Age from Offence Outcome
-0.002 M 8.110
X Y
5.492
*p<0.01
Figure A11. Predicting Perpetrator age (Y) from Offence Outcome (X) through the Use of
Drugs (M).
-0.083 M 0.092
X Y
5.486
*p<0.01
Figure A12. Predicting Perpetrator age (Y) from Offence Outcome (X) through the Use of
Alcohol (M).
0.181 -4.202*
M
X Y
6.241*
* p < 0.05; **p<0.01
Figure A13. Predicting Perpetrator age (Y) from Offence Outcome (X) through the Location
of Initial Contact (M).
207
27.876 M -0.000
X Y
5.490*
*p< 0.01
Figure A14. Predicting Perpetrator age (Y) from Offence Outcome (X) through Location of
Offence (M).
0.137 M -0.454
X Y
5.541
*p< 0.01
Figure A15. Predicting Perpetrator age (Y) from Offence Outcome (X) through Victim
Resistance (M).
Perpetrator Age from Victim Age
0.000 M 8.007
X Y
-0.039
Figure A16. Predicting Perpetrator age (Y) from Victim Age (X) through the Use of Drugs
(M).
0.001 M -0.304
X Y
-0.039
Figure A17. Predicting Perpetrator age (Y) from Victim Age (X) through the Use of Alcohol
(M).
208
-0.004 -3.225
M
X Y
-0.052
Figure A18. Predicting Perpetrator age (Y) from Victim Age (X) through the Location of
-0.543 0.002
M
X Y
-0.039
Figure A19. Predicting Perpetrator age (Y) from Victim Age (X) through the Location of
Offence (M).
0.00 M -0.336
X Y
-0.038
Figure A20. Predicting Perpetrator age (Y) from Victim Age (X) through Victim Resistance
(M).
Perpetrator Age from Weapon Use
0.044 M 8.913
X Y
Model Summary: R2=0.048, F=1.704, p=0.190
-3.163
Figure A21. Predicting Perpetrator age (Y) from Weapon Being Used (X) through Use of
Drugs (M).
-0.188 M -0.757
X Y
-2.913
Figure A22. Predicting Perpetrator age (Y) from Weapon Being Used (X) through Use of
Alcohol (M).
209
-0.021 -2.968
M
X Y
-2.833
Figure A23. Predicting Perpetrator age (Y) from Weapon Being Used (X) through the
Location of Initial Contact (M).
-18.646 M 0.002
X Y
-2.739
Figure A24. Predicting Perpetrator age (Y) from Weapon Being Used (X) through the
-0.129 M -0.397
X Y
Model Summary: R2=0.023, F= 0.797, p=0.455
-2.822
Figure A25. Predicting Perpetrator age (Y) from Weapon Being Used (X) through Victim
Resistance (M).
Perpetrator Age from Sexual Penetration
0.025 M 8.049
X Y
-0.020
Figure A26. Predicting Perpetrator age (Y) from Sexual Penetration (X) through Use of
Drugs(M).
0.117 -0.494
M
X Y
0.240
p < 0.05
210
Figure A27. Predicting Perpetrator age (Y) from Sexual Penetration (X) through Use of
Alcohol (M).
0.024 -2.951
M
X Y
0.252
Figure A28. Predicting Perpetrator age (Y) from Sexual Penetration (X) through Location of
-6.871 M 0.002
X Y
0.199
Figure A30. Predicting Perpetrator age (Y) from Sexual Penetration (X) through Location of
Offence (M).
0.068 -0.374
M
X Y
0.208
Figure A40. Predicting Perpetrator age (Y) from Sexual Penetration (X) through Victim
Resistance (M).
211
Table A1. Indirect Effects of Level of Aggression on Perpetrator Age through Contextual
Variables (ab paths)
Bootstrapping
Product of
Drugs -0.184 0.245 -0.754 -0.516 0.824 -1.549 0.276 -1.549 0.255
Alcohol -0.020 0.170 -0.117 -0.511 0.496 -0.644 0.366 -0.644 0.366
Location of Initial
Contact -0.271 0.290 -0.935 -1.011 0.164 -1.095 0.105 -1.046 0.135
Location of
Offence -0.024 0.109 -0.218 -0.289 0.247 -0.249 0.277 -0.361 0.193
Table A2. Indirect Effects of Level of Injuries on Perpetrator Age through Contextual
Bootstrapping
Product of
Drugs 0.000 0.000 0.405 -0.002 0.000 -0.000 0.001 -0.000 0.001
Alcohol 0.163 0.241 0.678 -0.334 0.813 -0.185 1.017 -0.200 0.974
Location of Initial
Contact -0.017 0.282 -0.059 -0.622 0.646 -0.603 0.667 -0.568 0.730
Location of
Offence 0.012 0.168 -0.073 -0.315 0.442 -0.343 0.413 -0.398 0.358
Table A3. Indirect Effects of Offence Outcome on Perpetrator Age through Contextual
Bootstrapping
Product of
Drugs -0.013 0.326 -0.041 -2.091 0.124 -0.568 0.464 -0.441 0.464
Alcohol -0.008 0.164 -0.047 -0.654 0.597 -0.737 0.513 -0.687 0.555
Location of Initial
Contact -0.076 0.597 -1.277 -2.328 0.143 -2.633 0.057 -2.600 0.063
Location of
Offence -0.011 0.231 -0.048 -1.073 0.295 -0667 0.593 -0.370 1.633
212
Table A4. Indirect Effects of Victim Age on Perpetrator Age through Contextual Variables
(ab paths)
Bootstrapping
Product of
Drugs -0.000 0.009 -0.041 -0.045 0.005 -0.014 0.021 -0.010 0.021
Alcohol -0.000 0.003 -0.137 -0.016 0.015 -0.022 0.010 -0.021 0.010
Location of Initial
Contact 0.012 0.012 0.948 -0.008 0.040 -0.004 0.048 -0.004 0.046
Location of
Offence -0.001 0.005 -0.198 -0.020 0.035 -0.021 0.033 -0.057 0.023
Table A5. Indirect Effects of a Weapon Being Used on Perpetrator Age through Contextual
Bootstrapping
Product of
Drugs 0.392 0.500 0.784 0.015 4.339 0.002 3.833 -0.031 2.914
Alcohol 0.142 0.402 0.353 -0.866 1.206 -0.506 1.686 -0.490 1.762
Location of Initial
Contact 0.062 0.402 0.154 -0.970 1.156 -0.821 1.304 -0.878 1.227
Location of
Offence -0.031 0.171 -0.184 -0.456 0.942 -0.486 0.927 -1.632 0.533
Victim Resistance 0.051 0.190 0.271 -0.700 1.093 -0.491 1.449 -0.569 1.288
Table A6. Indirect Effects of Sexual Penetration on Perpetrator Age through Contextual
Bootstrapping
Product of
Drugs 2.02 0.217 0.932 0.018 2.071 0.010 1.894 -0.012 1.461
Alcohol -0.058 0.251 -0.231 -0.614 0.607 -0.709 0.501 -0.753 0.470
Location of Initial
Contact -0.070 0.167 -0.422 -0.583 0.281 -0.737 0.191 -0.777 0.185
Location of
Offence -0.017 0.068 -0.245 -0.239 0.293 -0.161 0.393 -0.403 0.212
213
Table A7. Total Effects of Offence Variables on Perpetrator Age (c path)
Product of
Coefficients
B SE t p
Offence Outcome 5.479 1.963 2.791 0.007
Victim Age -0.039 0.054 -0.731 0.467
Weapon Used -2.771 2.444 -1.134 0.261
214
Relationship Status from Level of Aggression
0.009 M 0.460
X Y
0.034
Regression Summary: -2LL=131.002, Model LL=0.189, McFadden=0.001, Cox Snell=0.002, Nagelkrk=0.003
Figure A41. Predicting Relationship Status (Y) from Level of Aggression (X) through Use of
Drugs (M).
0.072 -0.414
M
X Y
0.068
Figure A42. Predicting Relationship Status (Y) from Level of Aggression (X) Use of Drugs
and Alcohol (M).
0.059 0.541
M
X Y
0.007
Regression Summary: -2LL=129.766, Model LL=1.426, McFadden=0. 011, Cox Snell=0.014, Nagelkrk=0.019
Figure A43. Predicting Relationship Status (Y) from Level of Aggression (X) through the
0.009 M 0.541
X Y
0.034
Regression Summary: -2LL=129.747, Model LL=1.445, McFadden=0.011 Cox Snell=0.014, Nagelkrk=0.019
Figure A44. Predicting Relationship Status (Y) from Level of Aggression (X) through the
215
0.216 0.019
M
X Y
0.033
Figure A45. Predicting Relationship Status (Y) from Level of Aggression (X) through Victim
Resistance (M).
Relationship Status from Level of Injury
0.000 M 0.429
X Y
0.000
Figure A46. Predicting Relationship Status (Y) from Level of Injury (X) through Use of
Drugs (M).
0.077 M -0.234
X Y
-0.252
Figure A47. Predicting Relationship Status (Y) from Level of Injury (X) through Use of
Alcohol (M).
0.030 M 0.771
X Y
-0.296
Figure A48. Predicting Relationship Status (Y) from Level of Injury (X) through the Location
216
0.013 1.294
M
X Y
-0.290
Figure A49. Predicting Relationship Status (Y) from Level of Injury (X) through the Location
of Offence (M).
-0.000 -0.118
M
X Y
-0.271
Figure A50. Predicting Relationship Status (Y) from Level of Injury (X) through Victim
Resistance (M).
Relationship Status from Offence Outcome
0.021 M 0.360
X Y
0.905*
*p < 0.05
Figure A51. Predicting Relationship Status (Y) from Offence Outcome (Y) through Use of
Drugs (M).
-0.167 M -0.269
X Y
0.870*
*p < 0.05
Figure A52. Predicting Relationship Status (Y) from Offence Outcome (Y) through Use of
Alcohol (M).
217
0.196* 0.377
M
X Y
0.845*
*p < 0.05
Figure A53 Predicting Relationship Status (Y) from Offence Outcome (Y) through the
0.154 M 0.415
X Y
0.856*
*p < 0.05
Figure A54. Predicting Relationship Status (Y) from Offence Outcome (Y) through the
0.047 0.018
M
X Y
0.911*
*p < 0.05
Figure A55. Predicting Relationship Status (Y) from Offence Outcome (Y) through Victim
Resistance (M).
Relationship Status from Victim Age
-0.001 M 0.445
X Y
-0.004
Figure A56. Predicting Relationship Status (Y) from Victim Age (X) through Use of Drugs
(M).
218
-0.001 -0.403
M
X Y
-0.001
Figure A57. Predicting Relationship Status (Y) from Victim Age (X) through Use of Alcohol
(M).
-0.004 0.525
M
X Y
-0.002
Figure A58. Predicting Relationship Status (Y) from Victim Age (X) through the Location of
-0.007* M 0.530
X Y
-0.001
*p < 0.01
Figure A59. Predicting Relationship Status (Y) from Victim Age (X) through the Location of
Offence (M).
0.001 0.023
M
X Y
-0.005
Figure A60. Predicting Relationship Status (Y) from Victim Age (X) through Victim
Resistance (M).
219
Relationship Status from Weapon Being Used
0.017 M 0.498
X Y
-0.420
Figure A61. Predicting Relationship Status (Y) from Weapon Used (X) through Use of Drugs
(M).
-0.001 -0.401
M
X Y
-0.415
Figure A62. Predicting Relationship Status (Y) from Weapon Used (X) through Use of
Alcohol (M).
-0.022 0.539
M
X Y
-0.406
Figure A63. Predicting Relationship Status (Y) from Weapon Used (X) through the Location
-0.216 M 0.496
X Y
-0.311
Figure A64. Predicting Relationship Status (Y) from Weapon Used (X) through the Location
of Offence (M).
220
0.158 0.027
M
X Y
-0.416
Figure A65. Predicting Relationship Status (Y) from Weapon Used (X) through Victim
Resistance (M).
Relationship Status from Sexual Penetration

0.015 M 0.560
X Y
-0.214
Regression Summary: -2LL=129.957, Model LL=1.234, McFadden=0.009, Cox Snell=0.044012 Nagelkrk=0.017
Figure A66. Predicting Relationship Status (Y) from Sexual Penetration (X) through Use of
Drugs (M).
0.125* -0.317
M
X Y
-0.175
*p < 0.05
Figure A67. Predicting Relationship Status (Y) from Sexual Penetration (X) through Use of
Alcohol (M).
0.033 0.583
M
X Y
-0.237
Figure A68. Predicting Relationship Status (Y) from Sexual Penetration (X) through the
221
-0.044 0.510
M
X Y
-0.195
Figure A69. Predicting Relationship Status (Y) from Sexual Penetration (X) through the
0.053 M 0.027
X Y
-0.216
Figure A70. Predicting Relationship Status (Y) from Sexual Penetration (X) through Victim
Resistance (M).
222
Table A8. Indirect Effects of Level of Aggression on Relationship Status through Contextual
Bootstrapping

CI
Drugs 0.004 0.291 -0.001 0.936 -0.010 0.320 -0.012 0.167
Alcohol -0.030 0.048 -0.148 0.044 -0.192 0.025 -0.182 0.026
Location of Initial
0.032 0.049 -0.043 0.159 -0.023 0.195 0.027 0.190
Contact
Location of Offence 0.005 0.052 -0.075 0.113 -0.060 0.129 -0.063 0.126
Table A9. Indirect Effects of Level of Injury on Relationship Status through Contextual
Bootstrapping

CI
Drugs 0.000 0.001 -0.001 0.002 -0.000 0.002 -0.001 0.001
Alcohol -0.018 0.061 -0.164 0.099 -0.216 0.063 -0.212 0.064
Location of Initial
0.023 0.132 -0.109 0.140 -0.047 0.230 -0.048 0.230
Contact
Table A10. Indirect Effects of Offence Outcome on Relationship Status through Contextual
Bootstrapping

CI
Drugs 0.007 0.395 -0.867 1.018 -0.236 1.618 -0.568 1.157
Alcohol 0.045 0.096 -0.124 0.266 -0.083 0.322 -0.084 0.320
Location of Initial
0.074 0.124 -0.111 0.387 -0.088 0.442 -0.095 0.422
Contact
223
Table A11. Indirect Effects of Victim Age on Relationship Status through Contextual
Bootstrapping

CI
Drugs -0.000 0.012 -0.039 0.000 -0.021 0.001 -0.016 0.002
Alcohol 0.000 0.002 -0.003 0.004 -0.002 0.005 -0.002 0.005
Location of Initial
-0.002 0.003 -0.010 0.002 -0.011 0.001 -0.010 0.001
Contact
Location of Offence -0.004 0.004 -0.012 0.003 -0.013 0.003 -0.012 0.003
Table A12. Indirect Effects of Weapon Used on Relationship Status through Contextual
Bootstrapping

CI
Drugs 0.008 0.664 -0.694 2.155 -0.868 1.879 -1.340 1.129
Alcohol 0.000 0.085 -0.183 0.184 -0.190 0.174 -0.197 0.169
Location of Initial
-0.012 0.097 -0.247 0.172 -0.296 0.140 -0.271 0.154
Contact
Table A13. Indirect Effects of Sexual Penetration on Relationship Status through Contextual
Bootstrapping

CI
Drugs 0.008 0.362 -0.279 1.200 -0.306 1.165 -0.544 0.786
Alcohol -0.040 0.065 -0.186 0.076 -0.201 0.065 -0.199 0.067
Location of Initial
0.019 0.045 -0.052 0.135 -0.029 0.185 -0.028 0.186
Contact
224
Table A14. Total Effects of Offence Variables on Relationship Status (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.912 0.431 2.116 0.034 4.476
Victim Age -0.005 0.011 -0.426 0.670 0.181
Weapon Used -0.411 0.520 -0.789 0.429 0.624
225
Any Previous Convictions from Level of Aggression
0.009 M -0.171
X Y
0.474
Figure A71. Predicting Any Previous Convictions (Y) from Level of Aggression (X) through
Use of Drugs (M).
0.072 M 0.395
X Y
0.449
Use of Alcohol (M).
0.059 M 0.432
X Y
0.451
the Location of Initial Contact (M).
0.009 M 0.829
X Y
0.482
the Location of Offence (M).
226
0.216 0.241
M
X Y
0.430
Victim Resistance (M).
Any Previous Convictions from Level of Injury
0.000 M -0.218
X Y
0.000
Figure A76. Predicting Any Previous Convictions (Y) from Level of Injury (X) through Use
of Drugs (M).
0.077 M -0.007
X Y
0.549*
*p<0.05
Figure A77. Predicting Any Previous Convictions (Y) from Level of Injury (X) through Use
of Alcohol (M).
0.030 M 0.011
X Y
0.548*
* p<0.05
Figure A78. Predicting Any Previous Convictions (Y) from Level of Injury (X) through the
227
0.013 1.647
M
X Y
0.548*
* p<0.05
Figure A79. Predicting Any Previous Convictions (Y) from Level of Injury (X) through the
-0.000 -0.057
M
X Y
0.551*
* p<0.05
Figure A80. Predicting Any Previous Convictions (Y) from Level of Injury (X) through
Any Previous Convictions from Offence Outcome
0.021 M -0.085
X Y
0.016
Figure A81. Predicting Any Previous Convictions (Y) from Offence Outcome (X) through
Use of Drugs (M).
-0.167 M 0.503
X Y
0.097
Use of Alcohol (M).
228
0.196 0.536
M
X Y
-0.086
0.154 M 0.827
X Y
-0.103
0.047 M 0.273
X Y
-0.007
Any Previous Convictions from Victim Age
-0.001 M 0.087
X Y
0.059*
*p<0.05
Figure A86. Predicting Any Previous Convictions (Y) from Victim Age (X) through Use of
Drugs (M).
229
-0.001 0.550
M
X Y
0.060*
*p<0.05
Figure A87. Predicting Any Previous Convictions (Y) from Victim Age (X) through Use of
Alcohol (M).
-0.004 M 0.786
X Y
0.062*
*p<0.01
Figure A88. Predicting Any Previous Convictions (Y) from Victim Age (X) through the
-0.007** M 1.345
X Y
0.071*
*p<0.05; ** p<0.01
Figure A89. Predicting Any Previous Convictions (Y) from Victim Age (X) through the
0.001 0.314
M
X Y
0.064*
* p<0.05
Figure A90. Predicting Any Previous Convictions (Y) from Victim Age (X) through Victim
Resistance (M).
230
Any Previous Convictions from Weapon Being Used
0.017 M -0.095
X Y
0.176
Figure A91. Predicting Any Previous Convictions (Y) from Weapon Used (X) through Use of
Drugs (M).
-0.001 M 0.488
X Y
0.176
Figure A92. Predicting Any Previous Convictions (Y) from Weapon Used (X) through Use of
Alcohol (M).
-0.022 M 0.522
X Y
0.187
Figure A93. Predicting Any Previous Convictions (Y) from Weapon Used (X) through the
-0.216 M 0.859
X Y
0.342
Figure A94. Predicting Any Previous Convictions (Y) from Weapon Used (X) through the
231
0.158 0.272
M
X Y
0.141
Figure A95. Predicting Any Previous Convictions (Y) from Weapon Used (X) through
Any Previous Convictions from Sexual Penetration

0.015 M -0.015
X Y
-0.176
Figure A96. Predicting Any Previous Convictions (Y) from Sexual Penetration (X) through
Use of Drugs (M).
0.125* 0.632
M
X Y
-0.263
*p <0.05
Use of Alcohol (M).
0.033 M 0.552
X Y
-0.197
232
-0.044 0.785
M
X Y
-0.149
0.053 M 0.282
X Y
-0.198
233
Table A15. Indirect Effects of Level of Aggression on Any Previous Convictions through
Bootstrapping

CI
Drugs -0.001 0.237 -0.584 0.550 -0.349 0.706 -0.184 1.182
Alcohol 0.028 0.054 -0.055 0.163 -0.032 0.213 -0.035 0.203
Location of Initial
0.026 0.087 -0.060 0.156 -0.030 0.228 -0.035 0.199
Contact
Table A16. Indirect Effects of Level of Injury on Any Previous Convictions through
Bootstrapping

CI
Drugs 0.000 0.001 -0.001 0.002 -0.001 0.002 -0.002 0.001
Alcohol -0.001 0.093 -0.154 0.131 -0.151 0.133 -0.150 0.134
Location of Initial
0.000 0.207 -0.081 0.173 -0.094 0.142 -0.105 0.121
Contact
Table A17. Indirect Effects of Offence Outcome on Any Previous Convictions through
Bootstrapping

CI
Drugs -0.002 0.392 -0.915 0.932 -0.652 1.157 -1.228 0.825
Alcohol -0.084 0.115 -0.360 0.080 -0.438 0.048 -0.429 0.053
Location of Initial
0.105 0.193 -0.097 0.450 -0.071 0.529 -0.081 0.492
Contact
234
Table A18. Indirect Effects of Victim Age on Any Previous Convictions through Contextual
Bootstrapping

CI
Drugs 0.000 0.010 -0.030 0.014 -0.027 0.017 -0.017 0.043
Alcohol -0.000 0.002 -0.005 0.004 -0.006 0.003 -0.006 0.003
Location of Initial
-0.004 0.006 -0.013 0.001 -0.014 0.001 -0.014 0.001
Contact
Location of Offence -0.010 0.010 -0.023 -0.001 -0.023 -0.001 -0.023 -0.001
Table A19. Indirect Effects of Weapon Used on Any Previous Convictions through
Bootstrapping

CI
Drugs -0.002 0.679 -0.681 2.161 -1.046 1.326 -2.551 0.713
Alcohol -0.000 0.107 -0.256 0.210 -0.229 0.222 -0.218 0.229
Location of Initial
-0.011 0.164 -0.246 0.203 -0.324 0.135 -0.297 0.151
Contact
Table A20. Indirect Effects of Sexual Penetration on Any Previous Convictions through
Bootstrapping

CI
Drugs -0.000 0.351 -0.259 1.181 -0.378 0.742 -0.665 0.430
Alcohol 0.079 0.098 -0.045 0.296 -0.033 0.326 -0.041 0.304
Location of Initial
0.018 0.067 -0.065 0.153 -0.035 0.237 -0.035 0.230
Contact
235
Table A21. Total Effects of Offence Variables on Any Previous Convictions (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.014 0.482 0.028 0.977 0.001
Victim Age 0.059 0.024 2.491 0.013 6.207
Weapon Used 0.174 0.619 0.282 0.778 0.080
236
Lives Alone from Level of Aggression
0.009 M -17.768
X Y
-0.581*
*p <0.05
Figure A101. Predicting Lives Alone (Y) from Level of Aggression (X) through Use of
Drugs (M).
0.072 M -0.545
X Y
-0.555*
*p <0.05
Figure A102. Predicting Lives Alone (Y) from Level of Aggression (X) through Use of
Alcohol (M).
0.059 -0.385
M
X Y
-0.564*
*p<0.05
Figure A103. Predicting Lives Alone (Y) from Level of Aggression (X) through the Location
0.009 M 0.107
X Y
-0.584 *
*p<0.05
Figure A104. Predicting Lives Alone (Y) from Level of Aggression (X) through the Location
of Offence (M).
237
0.216 -0.166
M
X Y
-0.552*
*p<0.05
Figure A105. Predicting Lives Alone (Y) from Level of Aggression (X) through Victim
Resistance (M).
Lives Alone from Level of Injury
0.000 M -17.937
X Y
0.000
Figure A106. Predicting Lives Alone (Y) from Level of Injury (X) through Use of Drugs and
Alcohol (M).
-0.000 M -0.641
X Y
0.000
Figure A107. Predicting Lives Alone (Y) from Level of Injury (X) through Use of Alcohol
(M).
0.000 -0.533
M
X Y
0.000
Figure A108. Predicting Lives Alone (Y) from Level of Injury (X) through the Location of
238
0.000 -0.533
M
X Y
0.000
Figure A109. Predicting Lives Alone (Y) from Level of Injury (X) through the Location of
Offence (M).
0.000 0.023
M
X Y
0.000
Figure A110. Predicting Lives Alone (Y) from Level of Injury (X) through Victim Resistance
(M).
Lives Alone from Offence Outcome
0.021 M -17.827
X Y
-0.119
Figure A111. Predicting Lives Alone (Y) from Offence Outcome (X) through Use of Drugs
(M).
-0.167 M -0.696
X Y
-0.260
Figure A112. Predicting Lives Alone (Y) from Offence Outcome (X) through Use of Alcohol
(M).
239
0.196* -0.484
M
X Y
-0.056
*p < 0.05
Figure A113. Predicting Lives Alone (Y) from Offence Outcome (X) through the Location of
0.154 0.119
M
X Y
-0.163
Figure A114. Predicting Lives Alone (Y) from Offence Outcome (X) through the Location of
Offence (M).
0.047 -0.208
M
X Y
-0.131
Figure A115. Predicting Lives Alone (Y) from Offence Outcome (X) through Victim
Resistance (M).
Lives Alone from Victim Age
-0.001 M -17.973
X Y
-0.037
Figure A116. Predicting Lives Alone (Y) from Victim Age (X) through Use of Drugs (M).
240
-0.001 -0.690
M
X Y
-0.036
Figure A117. Predicting Lives Alone (Y) from Victim Age (X) through Use of Alcohol (M).
-0.004 -0.685
M
X Y
-0.038
Figure A118. Predicting Lives Alone (Y) from Victim Age (X) through the Location of
-0.007* -0.209
M
X Y
-0.037
*p < 0.01
Figure A119. Predicting Lives Alone (Y) from Victim Age (X) through the Location of
Offence (M).
0.001 -0.221
M
X Y
-0.037
Figure A120. Predicting Lives Alone (Y) from Victim Age (X) through Victim Resistance
(M).
241
Lives Alone from Weapon Being Used
0.017 M -17.851
X Y
0.134
Figure A121. Predicting Lives Alone (Y) from Weapon Used (X) through Use of Drugs (M).
-0.001 M -0.652
X Y
0.112
Figure A122. Predicting Lives Alone (Y) from Weapon Used (X) through Use of Alcohol
(M).
-0.022 -0.494
M
X Y
0.101
Figure A123. Predicting Lives Alone (Y) from Weapon Used (X) through the Location of
-0.216 0.113
M
X Y
0.135
Figure A124. Predicting Lives Alone (Y) from Weapon Used (X) through the Location of
Offence (M).
242
0.158 -0.211
M
X Y
0.14
Figure A125. Predicting Lives Alone (Y) from Weapon Used (X) through Victim Resistance
(M).
Lives Alone from Sexual Penetration
0.015 M -17.722
X Y
-0.519
Figure A126. Predicting Lives Alone (Y) from Sexual Penetration (X) through Use of Drugs
(M).
0.125 -0.514
M
X Y
-0.476
Figure A127. Predicting Lives Alone (Y) from Sexual Penetration (X) through Use of
Alcohol (M).
0.033 -0.458
M
X Y
-0.516
Figure A128. Predicting Lives Alone (Y) from Sexual Penetration (X) through the Location
243
-0.044 0.026
M
X Y
-0.525
Figure A129. Predicting Lives Alone (Y) from Sexual Penetration (X) through the Location
of Offence (M).
0.053 -0.201
M
X Y
-0.517
Figure A130. Predicting Lives Alone (Y) from Sexual Penetration (X) through Victim
Resistance (M).
244
Table A22. Indirect Effects of Level of Aggression on Lives Alone through Contextual
Bootstrapping

CI
Drugs -0.151 0.402 -0.884 0.688 -0.880 0.696 -0.816 0.807
Alcohol -0.039 0.092 -0.229 0.046 -0.301 0.027 -0.273 0.032
Location of Initial
-0.023 0.141 -0.185 0.069 -0.247 0.037 -0.223 0.044
Contact
Victim Resistance -0.036 0.070 -0.215 0.065 -0.282 0.038 -0.269 0.041
Table A23. Indirect Effects of Level of Injury on Lives Alone through Contextual Variables
(ab paths)
Bootstrapping

CI
Drugs -0.001 0.001 -0.003 0.001 -0.003 0.000 -0.003 0.000
Alcohol 0.000 0.000 -0.000 0.000 -0.000 0.000 -0.000 0.000
Location of Initial
-0.000 0.000 -0.000 0.000 -0.001 0.000 -0.001 0.000
Contact
Table A24. Indirect Effects of Offence Outcome on Lives Alone through Contextual
Bootstrapping

CI
Drugs -0.367 0.673 -1.629 0.986 -1.603 1.022 -1.547 1.127
Alcohol 0.116 0.190 -0.076 0.487 -0.044 0.573 -0.053 0.539
Location of Initial
-0.095 0.259 -0.471 0.131 -0563 0.099 -0.515 0.112
Contact
245
Table A25. Indirect Effects of Victim Age on Lives Alone through Contextual Variables (ab
paths)
Bootstrapping

CI
Drugs 0.009 0.016 -0.021 0.041 -0.021 0.042 -0.020 0.043
Alcohol 0.000 0.004 -0.005 0.006 -0.004 0.007 -0.004 0.007
Location of Initial
0.003 0.008 -0.002 0.014 -0.001 0.014 -0.002 0.014
Contact
Table A26. Indirect Effects of Weapon Used on Lives Alone through Contextual Variables
(ab paths)
Bootstrapping

CI
Drugs -0.294 1.009 -2.594 1.241 -2.936 1.094 -4.533 0.898
Alcohol 0.000 0.182 -0.240 0.277 -0.240 0.277 -0.247 0.267
Location of Initial
0.012 0.190 -0.198 0.265 -0.141 0.372 -0159 0.320
Contact
Table A27. Indirect Effects of Sexual Penetration on Lives Alone through Contextual
Bootstrapping

CI
Drugs -0.264 0.498 -1.353 0.485 -1.585 0.436 -2.298 0.345
Alcohol -0.064 0.179 -0.287 0.071 -0.318 0.060 -0.297 0.067
Location of Initial
-0.015 0.115 -0.160 0.064 -0.272 0.035 -0.252 0.036
Contact
246
Table A28. Total Effects of Offence Variables on Lives Alone (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome -0.145 0.511 -0.283 0.777 0.080
Victim Age -0.036 0.020 -1.836 0.066 3.369
Weapon Used 0.110 0.628 0.176 0.861 0.031
247
Employed from Level of Aggression
0.009 M -19.728
X Y
-0.248
Figure A131. Predicting Being Employed (Y) from Level of Aggression (X) through Use of
Drugs (M).
0.072 M 0.035
X Y
-0.258
Figure A132. Predicting Being Employed (Y) from Level of Aggression (X) through Use of
Alcohol (M).
0.059 M 0.709
X Y
-0.302
Figure A133. Predicting Being Employed (Y) from Level of Aggression (X) through the
0.009 M 0.656
X Y
-0.267
Figure A134. Predicting Being Employed (Y) from Level of Aggression (X) through the
248
0.216 -0.037
M
X Y
-0.247
Figure A135. Predicting Being Employed (Y) from Level of Aggression (X) through Victim
Resistance (M).
Being Employed from Level of Injury
0.000 M -19.793
X Y
0.000
Figure A136. Predicting Being Employed (Y) from Level of Injury (X) through Use of Drugs
(M).
0.077 M -0.423
X Y
0.140
Figure A137. Predicting Being Employed (Y) from Level of Injury (X) through Use of
Alcohol (M).
0.030 M 1.353
X Y
0.074
Figure A138. Predicting Being Employed (Y) from Level of Injury (X) through the Location
249
0.013 0.815
M
X Y
0.097
Figure A139. Predicting Being Employed (Y) from Level of Injury (X) through the Location
of Offence (M).
-0.000 M -0.170
X Y
0.108
Figure A140. Predicting Being Employed (Y) from Level of Injury (X) through Victim
Resistance (M).
Being Employed from Offence Outcome
0.021 M -19.757
X Y
0.076
Figure A141. Predicting Being Employed (Y) from Offence Outcome (X) through Use of
Drugs (M).
-0.167 -0.021
M
X Y
0.014
Figure A142. Predicting Being Employed (Y) from Offence Outcome (X) through Use of
Alcohol (M).
250
0.196 0.662
M
X Y
-0.109
Figure A143. Predicting Being Employed (Y) from Offence Outcome (X) through the
0.154 M 0.652
X Y
-0.081
Figure A144. Predicting Being Employed (Y) from Offence Outcome (X) through the
0.047 M -0.057
X Y
0.020
Figure A145. Predicting Being Employed (Y) from Offence Outcome (X) through Victim
Resistance (M).
Being Employed from Victim Age
-0.001 M -19.727
X Y
0.005
Figure A146. Predicting Being Employed (Y) from Victim Age (X) through Use of Drugs
(M).
251
-0.001 -0.019
M
X Y
0.006
Figure A147. Predicting Being Employed (Y) from Victim Age (X) through Use of Alcohol
(M).
-0.004 M 0.702
X Y
0.009
Figure A148. Predicting Being Employed (Y) from Victim Age (X) through the Location of
-0.007* M 0.762
X Y
0.011
* p<0.01
Figure A149. Predicting Being Employed (Y) from Victim Age (X) through the Location of
Offence (M).
0.001 M -0.058
X Y
0.006
Figure A150. Predicting Being Employed (Y) from Victim Age (X) through Victim
Resistance (M).
252
Being Employed from Weapon Being Used
0.017 M -19.740
X Y
-0.113
Figure A151. Predicting Being Employed (Y) from Weapon Used (X) through Use of Drugs
(M).
-0.001 -0.023
M
X Y
-0.148
Figure A152. Predicting Being Employed (Y) from Weapon Used (X) through Use of
Alcohol (M).
0.638
-0.022 M
X Y
-0.137
Figure A153. Predicting Being Employed (Y) from Weapon Used (X) through the Location
-0.216 M 0.637
X Y
-0.016
Figure A154. Predicting Being Employed (Y) from Weapon Used (X) through the Location
of Offence (M).
253
0.158 -0.055
M
X Y
-0.139
Figure A155. Predicting Being Employed (Y) from Weapon Used (X) through Victim
Resistance (M).
Being Employed from Sexual Penetration

0.015 M -19.721
X Y
-0.085
Figure A156. Predicting Being Employed (Y) from Sexual Penetration (Y) through Use of
Drugs (M).
0.125* 0.032
M
X Y
-0.119
p < 0.05
Figure A157. Predicting Being Employed (Y) from Sexual Penetration (Y) through Use of
Alcohol (M).
0.033 M 0.661
X Y
-0.140
Figure A158. Predicting Being Employed (Y) from Sexual Penetration (Y) through the
254
-0.044 0.624
M
X Y
-0.090
Figure A159. Predicting Being Employed (Y) from Sexual Penetration (Y) through the
0.053 M -0.054
X Y
-0.112
Figure A160. Predicting Being Employed (Y) from Sexual Penetration (Y) through Victim
Resistance (M).
255
Table A29. Indirect Effects of Level of Aggression on Being Employed through Contextual
Bootstrapping

CI
Drugs -0.167 0.445 -0.980 0.765 -0.935 0.833 -0.874 0.964
Alcohol 0.003 0.041 -0.081 0.097 -0.069 0.115 -0.070 0.113
Location of Initial
0.042 0.051 -0.040 0.168 -0.021 0.203 -0.023 0.199
Contact
Table A30. Indirect Effects of Level of Injury on Being Employed through Contextual
Bootstrapping

CI
Drugs -0.001 0.001 -0.003 0.001 -0.003 0.000 -0.003 0.000
Alcohol -0.033 0.066 -0.195 0.081 -0.264 0.042 -0.249 0.047
Location of Initial
0.041 0.204 -0.229 0.237 -0.119 0.535 -0.120 0.525
Contact
Table A31. Indirect Effects of Offence Outcome on Being Employed through Contextual
Bootstrapping

CI
Drugs -0.406 0.751 -1.860 1.141 -1.795 1.222 -1.714 1.333
Alcohol 0.004 0.085 -0.180 0.185 -0163 0.204 -0.163 0.204
Location of Initial
0.130 0.130 -0.047 0.454 -0.027 0.524 -0.031 0.491
Contact
256
Table A32. Indirect Effects of Victim Age on Being Employed through Contextual Variables
(ab paths)
Bootstrapping

CI
Drugs 0.010 0.017 -0.025 0.043 -0.025 0.044 -0.023 0.045
Alcohol 0.000 0.001 -0.003 0.003 -0.002 0.003 -0.002 0.003
Location of Initial
-0.003 0.003 -0.010 0.001 -0.011 0.000 -0.010 0.001
Contact
Table A33. Indirect Effects of Weapon Used on Being Employed through Contextual
Bootstrapping

CI
Drugs -0.326 1.117 -2.821 1.375 -3.180 1.233 -4.751 1.014
Alcohol 0.000 0.057 -0.116 0.126 -0.116 0.125 -0.123 0.117
Location of Initial
-0.014 0.102 -0.259 0.181 -0.291 0.156 -0.270 0.169
Contact
Table A34. Indirect Effects of Sexual Penetration on Being Employed through Contextual
Bootstrapping

CI
Drugs -0.294 0.581 -1.612 0.499 -1.705 0.478 -2.615 0.400
Alcohol 0.004 0.058 -0.113 0.122 -0.108 0.129 -0.110 0.127
Location of Initial
0.022 0.048 -0.062 0.140 -0.039 0.175 -0.038 0.177
Contact
257
Table A35. Total Effects of Offence Variables on Being Employed (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.017 0.419 0.041 0.967 0.002
Victim Age 0.006 0.011 0.527 0.599 0.277
Weapon Used -0.148 0.517 -0.286 0.775 0.082
258
Appendix E: Content Dictionary for Implicit Theories Coding
The five identified Implicit Theories for sexual aggressors against adult females are listed
below, with identifying features and examples of quotations taken from the various
interviews undertaken in previous studies (e.g., Beech et al., 2005).
1) Women are Unknowable Dangerous

 Women are inherently different from men; these differences cannot be understood
readily by men
 Women are inherently deceptive
 Offender will not seek intimacy with woman; will keep sex impersonal and
relationship superficial
 Women know that their own desires/needs are incompatible with men and
therefore they do not communicate these desires directly but present them in a
disguised manner
“nice girls or whores”

“when women say no they really mean yes”
“just playing hard to get”
“women are sly and manipulating”
“really want a man to force her to have sex so she doesn’t seem loose”
Victim focused
2) Women as Sex Objects

 Women exist in a constant state of sexual reception
 Women constantly desire sex
 They should always be reception/available to meet the sexual needs of a man
 Men misattribute sexual intent to women’s nonsexual behaviour
 Women do not deliberately deceive men
 If a guy spends time and money on a girl its than expected and known she has to
sleep with him if he so wants; moral obligation
“a woman can enjoy sex even when its forced upon her”
“rape is generally a misinterpretation of sexual cues”
“a woman should feel guilty following a rape”
“a raped woman is a responsible victim not an innocent one”
“only women who are physically beaten should feel justified in reporting a rape”
“many women have an unconscious wish to be rape and then unconsciously set up a situation in
which they are likely to be attacked”
“a woman who changed her mind afterwards”
Victim focused
3) Male Sex Drive is Uncontrollable
259
 Men who rape attribute the causes of their offending to external factors
 Located in victim or in other features of the environment (i.e. alcohol)
 Male sex drive is hard to control and women play a key role in it loss of control
 Sexual build up can lead to aggressive outbursts
“women falsely accuse men of rape to protect their reputation”

“rape is a way to get back at a former lover”
“women can prevent being raped if they really want to”
“sexual offence occurred because she didn’t understand me”
“she was asking for it”
“nice women don’t get raped”
General theory; any potential perpetrator
4) Entitlement
 Men should have their needs, including their sexual needs meet on demand
 Men are inherently superior to women
 Women are sexually naïve and psychologically immature therefore men are
entitled to control women’s sexuality and to determine what a woman really wants
 Any man is entitled to punish a women for unsuitable conduct and the punishment
can be rape if he wants sex
 Some people are superior to and more important than others
“being a whore or acting too good for a man justifies rape”

“if would do some women good to be raped”
“rape puts women in their place”
“a wife should always be sexual when required”
“women are there to meet men’s sexual needs regardless of their own”
“men rape because women reject them”
“a spouse owes their partner sex no matter the circumstance”
Male specific
5) Dangerous World
 The world is inherently a hostile and uncaring place where by default others are
out to harm, exploit and degrade and deceive in order to promote their own
interests
 Often works in tandem with entitlement to justify and support exploitative and
harmful behaviour towards others
 Perceive threats where evidence is absent or ambiguous
 Supports hostile behaviour towards others a s pre-emptive action to prevent
inevitable harm to the self
 Necessary to fight back and achieve dominance and control over other people
 If women are perceived as threats and in need of retribution they become victims
of sexual abuse
“it’s a dog eat god world”

“she would have done the same to me if I hadn’t got to her first”
“lots of people are out to get you”
260
“control or be controlled”
“I did it to get revenge on her and her mother”
“I had to teach her a lesson”
“she had no right to question my authority”
General theory; predict others behaviour
261
Appendix F: Figures and Tables of the Implicit Theories Mediation Analyses
0.219* M -0.529
X Y
-0.071
*p<0.05
Figure C1. Predicting Perpetrator age (Y) from Aggression (X) through Dangerous World IT
(M).
-0.043 M 3.085
X Y
-0.054
Figure C2. Predicting Perpetrator age (Y) from Aggression (X) through Women as Sex
Objects IT (M).
-0.073 M -0.769
X Y
-0.243
Figure C3. Predicting Perpetrator age (Y) from Aggression (X) through Male Sex Drive
Uncontrollable IT (M).
-0.069 M -5.010
X Y
-0.531
Figure C4. Predicting Perpetrator age (Y) from Aggression (X) through Entitlement IT (M).
262
0.188* M -0.226
X Y
-2.021
Figure C5. Predicting Perpetrator age (X) from Level of Injury (Y) through Dangerous World
IT (M).
-0.004 M 4.178
X Y
-2.048*
* p=0.05
Figure C6. Predicting Perpetrator age (X) from Level of Injury (Y) through Women as Sex
Objects IT (M).
-0.105 M -1.071
X Y
-2.177*
* p < 0.05
Figure C7. Predicting Perpetrator age (X) from Level of Injury (Y) through Male Sex Drive
-0.034 M -5.756
X Y
-2.259*
*p <0.05
Figure C8. Predicting Perpetrator age (X) from Level of Injury (Y) through Entitlement IT
(M).
263
-0.251 M 0.087
X Y
4.343
*p<0.01
Figure C9. Predicting Perpetrator age (X) from Offence Outcome (Y) through Dangerous
World IT (M).
0.072 2.616
M
X Y
4.132*
* p<0.05
Figure C10. Predicting Perpetrator age (X) from Offence Outcome (Y) through Women as
Sex Objects IT (M).
0.033 -0.895
M
X Y
4.351*
*p<0.05
Figure C11. Predicting Perpetrator age (X) from Offence Outcome (Y) through Male Sex
Drive Uncontrollable IT (M).
0.052 M -5.498
X Y
4.606
*p<0.05
Figure C12. Predicting Perpetrator age (X) from Offence Outcome (Y) through Entitlement
IT (M).
264
-0.001 M -0.561
X Y
-0.015
Figure C13. Predicting Perpetrator age (X) from Victim Age (Y) through Dangerous World
IT (M).
0.003 3.284
M
X Y
Model Summary: R2=0.029, F=0.942p=0.395
-0.026
Figure C14. Predicting Perpetrator age (X) from Victim Age (Y) through Women as Sex
Objects IT (M).
-0.004 -0.827
M
X Y
-0.018
Figure C15. Predicting Perpetrator age (X) from Victim Age (Y) through Male Sex Drive
-0.001 M -4.834
X Y
-0.019
Figure C16. Predicting Perpetrator age (X) from Victim Age (Y) through Entitlement IT (M).
265
0.460* M -0.055
X Y
-2.530
*p<0.05
Figure C17. Predicting Perpetrator age (X) from Weapon Being Used (Y) through Dangerous
World IT (M).
-0.009 M 3.067
X Y
-2.529
Figure C18. Predicting Perpetrator age (X) from Weapon Being Used (Y) through Women as
Sex Objects IT (M).
-0.310* M -1.540
X Y
-3.033
*p<0.05
Figure C19. Predicting Perpetrator age (X) from Weapon Being Used (Y) through Male Sex
0.047 M -4.524
X Y
-2.345
Figure C20. Predicting Perpetrator age (X) from Weapon Being Used (Y) through
Entitlement IT (M).
266
0.077 M -0.616
X Y
0.325
Figure C21. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Dangerous
World IT (M).
-0.015 M 3.129
X Y
0.325
Figure C22. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Women as
Sex Objects IT (M).
-0.040 M -0.669
X Y
0.251
Figure C23. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Male Sex
-0.043 M -4.724
X Y
0.073
Figure C24. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Entitlement
IT (M).
267
Table C1. Indirect Effects of Level of Aggression on Perpetrator Age through Implicit
Bootstrapping
Product of
Dangerous World -0.116 0.335 -0.347 -0.828 0.634 -0.832 0.632 -0.862 0.614
Women as Sex
Objects -0.133 0.228 -0.584 -0.854 0.496 -1.378 0.214 -1.455 0.199
Male Sex Drive
Uncontrollable 0.056 0.164 0.342 -0.406 0.615 -0.220 0.872 -0.244 0.819
Entitlement 0.344 0.329 1.047 -0.185 1.143 -0.116 1.689 -0.113 1.752
Table C2. Indirect Effects of Level of Injuries on Perpetrator Age through Implicit Variables
(ab paths)
Bootstrapping
Product of
Dangerous World -0.043 0.296 -0.144 -0.565 0.679 -0.565 0.680 -0.614 0.600
Women as Sex
Objects -0.157 0.197 -0.080 -0.552 0.640 -0.668 0.488 -0.853 0.401
Male Sex Drive
Uncontrollable 0.113 0.267 0.423 -0.517 0.643 -0.239 0.864 -0.251 0.854
Entitlement 0.195 0.258 0.755 -0.296 1.061 -0.189 1.364 -0.164 1.511
Table C3. Indirect Effects of Offence Outcome on Perpetrator Age through Implicit Variables
(ab paths)
Bootstrapping
Product of
Dangerous World -0.022 0.359 -0.061 -0.808 0.719 -0.950 0.609 -0.923 0.623
Women as Sex
Objects 0.189 0.304 0.623 -0.528 1.168 -0.235 1.674 -0.238 1.650
Male Sex Drive
Uncontrollable -0.030 0.121 -0.245 -0.484 0.466 -0.782 0.240 -0.768 0.248
Entitlement -0.285 0.418 -0.680 -1.380 0.456 -1.581 0.345 -1.560 0.355
268
Table C4. Indirect Effects of Victim Age on Perpetrator Age through Implicit Variables (ab
paths)
Bootstrapping
Product of
Dangerous World 0.001 0.003 0.188 -0.010 0.016 -0.011 0.015 -0.012 0.015
Women as Sex
Objects 0.011 0.012 0.939 -0.012 0.052 -0.005 0.071 -0.005 0.070
Male Sex Drive
Uncontrollable 0.003 0.008 0.373 -0.016 0.019 -0.009 0.030 -0.009 0.029
Entitlement 0.004 0.009 0.422 -0.011 0.020 -0.007 0.025 -0.006 0.027
Table C5. Indirect Effects of a Weapon Being Used on Perpetrator Age through Implicit
Bootstrapping
Product of
Dangerous World -0.025 0.697 -0.036 -1.330 1.154 -1.336 1.152 -1.370 1.129
Women as Sex
Objects -0.026 0.351 -0.075 -1.295 0.755 -1.232 0.823 -1.084 1.018
Male Sex Drive
Uncontrollable 0.477 0.694 0.687 -0.779 1.836 -0.614 2.074 -0.645 2.073
Entitlement -0.211 0.402 -0.524 -1.130 0.868 -1.745 0.442 -2.196 0.363
Table C6. Indirect Effects of Sexual Penetration on Perpetrator Age through Implicit
Bootstrapping
Product of
Dangerous World -0.047 0.122 -0.388 -0.348 0.384 -0.512 0.186 -0.651 0.158
Women as Sex
Objects -0.048 0.147 -0.328 -0.570 0.226 -0.590 0.216 -0.487 0.341
Male Sex Drive
Uncontrollable 0.027 0.090 0.295 -0.202 0.361 -0.135 0.460 -0.156 0.418
Entitlement 0.204 0.211 0.971 -0.033 0.510 -0.011 0.545 -0.010 0.597
269
Table C7. Total Effects of Offence Variables on Perpetrator Age (c path)
Product of
Coefficients
B SE t p
Offence Outcome 4.321 1.850 2.335 0.023
Victim Age -0.015 0.051 -0.286 0.776
Weapon Used -2.555 2.235 -1.143 0.257
270
Relationship Status from Level of Aggression Used
0.106 M -0.261
X Y
0.050
Regression Summary: -2LL=120.018, Model LL=0.550, McFadden=0.005, CoxSnell=0.006, Nagelkrk=0.008
Figure C25. Predicting Relationship Status (X) from Aggression (Y) through Dangerous
World IT.
-0.038 M 0.044
X Y
0.023
Figure C26. Predicting Relationship Status (X) from Aggression (Y) through Women as Sex
Objects IT.
-0.246
-0.015 M
X Y
0.017
Figure C27. Predicting Relationship Status (X) from Aggression (Y) through Male Sex Drive
-0.027 M 1.014
X Y
0.044
Regression Summary: -2LL=119.576, Model LL=991, McFadden=0.008, CoxSnell=0.011, Nagelkrk=0.015
Figure C28. Predicting Relationship Status (X) from Aggression (Y) through Entitlement IT
(M).
271
0.167* M -0.474
X Y
-0.191
*p<0.05
Figure C29. Predicting Relationship Status (X) from Level of Injury (Y) through Dangerous
World IT (M).
-0.028 -0.177
M
X Y
-0.273
Figure C30. Predicting Relationship Status (X) from Level of Injury (Y) through Women as
Sex Objects IT (M).
-0.090 -0.111
M
X Y
-0.279
Figure C31. Predicting Relationship Status (X) from Level of Injury (Y) through Male Sex
-0.008 M 18.70
X Y
-0.269
Figure C32. Predicting Relationship Status (X) from Level of Injury (Y) through Entitlement
IT (M).
272
-0.192 M -0.153
X Y
0.866
Figure C33. Predicting Relationship Status (X) from Offence Outcome (Y) through
Dangerous World IT (M)
0.074 M -0.051
X Y
0.898*
* p=0.05
Figure C34. Predicting Relationship Status (X) from Offence Outcome (Y) through Women
as Sex Objects IT (M).
0.023 M -0.282
X Y
0.904*
*p=0.05
Figure C35. Predicting Relationship Status (X) from Offence Outcome (Y) through Male Sex
0.008 M 1.009
X Y
0.896
Regression Summary: -2LL=115.662, Model LL=4.906, McFadden=0.041, CoxSnell=0.051, Nagelkrk=0.070; *p=0.05
Figure C36. Predicting Relationship Status (X) from Offence Outcome (Y) through
Entitlement IT (M).
273
-0.001 M -0.260
X Y
-0.008
Figure C37. Predicting Relationship Status (X) from Victim Age (Y) through Dangerous
World IT (M).
0.002 0.068
M
X Y
-0.008
Figure C38. Predicting Relationship Status (X) from Victim Age (Y) through Women as Sex
Objects IT (M).
-0.001 -0.270
M
X Y
-0.008
Figure C39. Predicting Relationship Status (X) from Victim Age (Y) through Male Sex Drive
-0.001 M 0.977
X Y
-0.007
Figure C40. Predicting Relationship Status (X) from Victim Age (Y) through Entitlement IT
(M).
274
Relationship Status from Weapon Use
0.311* M -0.201
X Y
-0.373
Regression Summary: -2LL=119.580, Model LL=0.988, McFadden=0.008 CoxSnell=0.011, Nagelkrk=0.015
*p=0.05
Figure C41. Predicting Relationship Status (X) from Weapon Being Used (Y) through
Dangerous World IT (M).
-0.082 M 0.006
X Y
-0.435
Women as Sex Objects IT (M).
-0.123 M -0.295
X Y
-0.474
Figure C43. Predicting Relationship Status (X) from Weapon Being Used (Y) through Male
Sex Drive Uncontrollable IT (M).
0.052 M 1.082
X Y
-0.489
Entitlement IT (M).
275
0.064 M -0.220
X Y
-0.211
Figure C45. Predicting Relationship Status (X) from Sexual Penetration (Y) through
-0.027 M 0.006
X Y
-0.224
Figure C46. Predicting Relationship Status (X) from Sexual Penetration (Y) through Women
-0.022 M -0.273
X Y
-0.230
Figure C47. Predicting Relationship Status (X) from Sexual Penetration (Y) through Male
-0.030 M 0.899
X Y
-0.202
Figure C48. Predicting Relationship Status (X) from Sexual Penetration (Y) through
Entitlement IT (M).
276
Table C8. Indirect Effects of Level of Aggression on Relationship Status through Implicit
Bootstrapping

CI
Dangerous World -0.028 0.057 -0.160 0.074 -0.197 0.051 -0.216 0.042
Women as Sex
-0.002 0.047 -0.079 0.093 -0.117 0.063 -0.110 0.068
Objects
Male Sex Drive
Uncontrollable 0.004 0.035 -0.068 0.081 -0.039 0.119 -0.039 0.119
Entitlement -0.027 0.343 -0.944 0.662 -1.461 0.364 -1.010 0.640
Table C9. Indirect Effects of Level of Injury on Relationship Status through Implicit
Bootstrapping

CI
Dangerous World -0.079 0.106 -0.328 0.096 -0.337 0.088 -0.354 0.081
Women as Sex
0.005 0.206 -0.554 0.146 -0.083 0.287 -0.087 0.250
Objects
Male Sex Drive
Uncontrollable 0.010 0.092 -0.140 0.180 -0.122 0.204 -0.123 0.201
Entitlement -0.149 0.595 -1.366 0.965 -1.391 0.953 -1.555 0.874
Table C10. Indirect Effects of Offence Outcome on Relationship Status through Implicit
Bootstrapping

CI
Dangerous World 0.029 0.099 -0.130 0.291 -0.104 0.332 -0.090 0.363
Women as Sex
-0.004 0.073 -0.182 0.121 -0.200 0.111 -0.200 0.111
Objects
Male Sex Drive
Uncontrollable -0.006 0.058 -0.127 0.119 -0.179 0.079 -0.165 0.083
Entitlement 0.008 0.583 -1.676 1.177 -0.688 1.818 -0.221 2.814
277
Table C11. Indirect Effects of Victim Age on Relationship Status through Implicit Variables
(ab paths)
Bootstrapping

CI
Dangerous World 0.000 0.002 -0.002 0.004 -0.002 0.005 -0.002 0.005
Women as Sex
0.000 0.003 -0.003 0.006 -0.003 0.005 -0.004 0.005
Objects
Male Sex Drive
Uncontrollable 0.000 0.001 -0.003 0.004 -0.001 0.005 -0.002 0.005
Entitlement -0.001 0.013 -0.042 0.001 -0.036 0.002 -0.027 0.026
Table C12. Indirect Effects of Weapon Used on Relationship Status through Implicit
Bootstrapping

CI
Dangerous World -0.062 0.154 -0.440 0.184 -0.581 0.116 -0.656 0.099
Women as Sex
-0.001 0.128 -0.203 0.127 -0.170 0.153 -0.156 0.165
Objects
Male Sex Drive
Uncontrollable 0.036 0.090 -0.108 0.272 -0.062 0.361 -0.069 0.347
Entitlement 0.056 1.051 -0.730 3.594 -0.699 3.720 -1.398 2.655
Table C13. Indirect Effects of Sexual Penetration on Relationship Status through Implicit
Bootstrapping

CI
Dangerous World -0.014 0.043 -0.093 0.093 -0.175 0.036 -0.220 0.027
Women as Sex
-0.000 0.037 -0.087 0.032 -0.052 0.050 -0.048 0.056
Objects
Male Sex Drive
Uncontrollable 0.006 0.031 -0.063 0.072 -0.030 0.107 -0.036 0.094
Entitlement -0.027 0.262 -0.843 0.025 -0.744 0.032 -0.572 0.332
278
Table C14. Total Effects of Offence Variables on Relationship Status (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.894 0.451 1.982 0.047 3.093
Victim Age -0.008 0.011 -0.686 0.493 0.471
Weapon Used -0.435 0.527 -0.827 0.408 0.684
279
Previous Convictions from Level of Aggression Used
0.106 M 0.005
X Y
0.233
Figure C49. Predicting Previous Convictions (X) from Aggression (Y) through Dangerous
World IT (M).
-0.038 0.133
M
X Y
0.238
Figure C50. Predicting Previous Convictions (X) from Aggression (Y) through Women as
Sex Objects IT (M).
-0.015 M -0.032
X Y
0.233
Figure C51. Predicting Previous Convictions (X) from Aggression (Y) through Male Sex
-0.027 0.328
M
X Y
0.241
Figure C52. Predicting Previous Convictions (X) from Aggression (Y) through Entitlement
IT (M).
280
Previous Convictions from Level of Injury
0.167* M -0.292
X Y
0.602*
*p<0.05
Figure C53. Predicting Previous Convictions (X) from Level of Injury (Y) through
-0.028 M -0.432
X Y
0.535*
* p<0.05
Figure C54. Predicting Previous Convictions (X) from Level of Injury (Y) through Women as
Sex Objects IT (M).
-0.090 0.977
M
X Y
0.638*
Regression Summary: -2LL=58.432, Model LL=6.653, McFadden=0.1.02, CoxSnell=0.112, Nagelkrk=0.163
*p<0.05
Figure C55. Predicting Previous Convictions (X) from Level of Injury (Y) through Male Sex
-0.008 M 0.227
X Y
0.551*
Regression Summary: -2LL=60.357, Model LL=4.728, McFadden=0.073, CoxSnell=0.081, Nagelkrk=0.118; *p<0.05
Figure C56. Predicting Previous Convictions (X) from Level of Injury (Y) through
Entitlement IT (M).
281
Previous Convictions from Offence Outcome
-0.192 M 0.057
X Y
0.047
Figure C57. Predicting Previous Convictions (X) from Offence Outcome (Y) through
0.074 M 0.092
X Y
0.030
Regression Summary: -2LL=94.601, Model LL=0.026, McFadden=0.000, CoxSnell=0.000 Nagelkrk=0.000
Figure C58. Predicting Previous Convictions (X) from Offence Outcome (Y) through Women
0.023 -0.044
M
X Y
0.037
Figure C59. Predicting Previous Convictions (X) from Offence Outcome (Y) through Male
0.008 M 0.250
X Y
0.035
Figure C60. Predicting Previous Convictions (X) from Offence Outcome (Y) through
Entitlement IT (M).
282
Previous Convictions from Victim Age
-0.001 M 0.080
X Y
0.052*
*p<0.05
Figure C61. Predicting Previous Convictions (X) from Victim Age (Y) through Dangerous
World IT (M).
0.002 0.128
M
X Y
0.053*
*p<0.05
Figure C62. Predicting Previous Convictions (X) from Victim Age (Y) through Women as
Sex Objects IT (M).
-0.001 M -0.030
X Y
0.052*
*p<0.05
Figure C63. Predicting Previous Convictions (X) from Victim Age (Y) through Male Sex
-0.001 M 0.283
X Y
0.052*
*p<0.05
Figure C64. Predicting Previous Convictions (X) from Victim Age (Y) through Entitlement
IT (M).
283
Previous Convictions from Weapon Use
0.311* M 0.063
X Y
-0.083
*p=0.05
Figure C65. Predicting Previous Convictions (X) from Weapon Being Used (Y) through
-0.082 M 0.090
X Y
-0.057
-0.123 M -0.049
X Y
-0.071
Male Sex Drive Uncontrollable IT (M).
0.052 M 0.262
X Y
-0.077
Entitlement IT (M).
284
Previous Convictions from Sexual Penetration
0.064 M 0.111
X Y
-0.283
Figure C69. Predicting Previous Convictions (X) from Sexual Penetration (Y) through
-0.027 M 0.046
X Y
-0.275
-0.022 M -0.071
X Y
-0.277
Regression Summary: -2LL=93.321, Model LL=1.307, McFadden=0.0.14, CoxSnell=0.014, Nagelkrk=0.022
Figure C71. Predicting Previous Convictions (X) from Sexual Penetration (Y) through Male
-0.030 0.099
M
X Y
-0.273
Entitlement IT (M).
285
Table C15. Indirect Effects of Level of Aggression on Previous Convictions through Implicit
Bootstrapping

CI
Dangerous World 0.001 0.059 -0.105 0.127 -0.105 0.125 -0.106 0.124
Women as Sex
-0.005 0.161 -0.140 0.090 -0.709 0.053 -0.216 0.062
Objects
Male Sex Drive
Uncontrollable 0.001 0.060 -0.071 0.090 -0.075 0.086 -0.070 0.092
Entitlement -0.009 0.504 -1.669 0.268 -1.460 0.369 -1.046 0.869
Table C16. Indirect Effects of Level of Injury on Previous Convictions through Implicit
Bootstrapping

CI
Dangerous World -0.049 0.117 -0.291 0.142 -0.288 0.143 -0.249 0.188
Women as Sex
0.012 0.208 -0.052 0.587 -0.044 0.741 -0.046 0.702
Objects
Male Sex Drive
Uncontrollable -0.088 0.273 -0.441 0.050 -0637 0.030 -0.488 0.043
Entitlement -0.002 0.366 -1.198 0.302 -0.444 0.800 -0.227 1.249
Table C17. Indirect Effects of Offence Outcome on Previous Convictions through Implicit
Bootstrapping

CI
Dangerous World -0.011 0.115 -0.306 0.165 -0.345 0.146 -0.337 0.150
Women as Sex
0.007 0.222 -0.187 0.198 -0.112 0.822 -0.122 0.325
Objects
Male Sex Drive
Uncontrollable -0.001 0.120 -0.113 0.151 -0.138 0.126 -0.153 0.116
Entitlement 0.002 0.583 -0.561 1.794 -1.446 0.989 -2.673 0.543
286
Table C18. Indirect Effects of Victim Age on Previous Convictions through Implicit
Bootstrapping

CI
Dangerous World -0.000 0.002 -0.004 0.003 -0.004 0.003 -0.004 0.003
Women as Sex
0.000 0.009 -0.005 0.007 -0.003 0.030 -0.004 0.010
Objects
Male Sex Drive
Uncontrollable 0.000 0.003 -0.004 0.004 -0.003 0.004 -0.003 0.004
Entitlement -0.000 0.010 -0.031 0.017 -0.034 0.014 -0.023 0.032
Table C19. Indirect Effects of Weapon Used on Previous Convictions through Implicit
Bootstrapping

CI
Dangerous World 0.020 0.171 -0.214 0.487 -0.199 0.530 -0.203 0.519
Women as Sex
-0.007 0.243 -0.248 0.191 -0.582 0.132 -0.380 0.145
Objects
Male Sex Drive
Uncontrollable 0.006 0.175 -0.208 0.244 -0.199 0.257 -0.187 0.272
Entitlement 0.014 0.695 -1.407 1.881 -0.280 3.701 -0.398 3.379
Table C20. Indirect Effects of Sexual Penetration on Previous Convictions through Implicit
Bootstrapping

CI
Dangerous World 0.007 0.050 -0.081 0.124 -0.049 0.185 -0.051 0.171
Women as Sex
-0.001 0.142 -0.118 0.055 -0.119 0.055 -0.094 0.069
Objects
Male Sex Drive
Uncontrollable 0.002 0.054 -0.070 0.085 -0.064 0.090 -0.066 0.086
Entitlement -0.003 0.265 -0.812 0.056 -0.767 0.063 -0.601 0.259
287
Table C21. Total Effects of Offence Variables on Previous Convictions (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.036 0.533 0.068 0.946 0.005
Victim Age 0.052 0.024 2.163 0.031 4.679
Weapon Used -0.065 0.633 -0.102 0.919 0.010
288
Lives Alone from Level of Aggression Used
0.106 M -0.488
X Y
-0.505
Figure C73. Predicting Lives Alone (X) from Aggression (Y) through Dangerous World IT
(M).
-0.038 M -0.009
X Y
-0.543
Figure C74. Predicting Lives Alone (X) from Aggression (Y) through Women as Sex Objects
IT (M).
-0.015 M 0.873
X Y
-0.550
Figure C75. Predicting Lives Alone (X) from Aggression (Y) through Male Sex Drive
-0.027 -17.980
M
X Y
-0.593
Figure C76. Predicting Lives Alone (X) from Aggression (Y) through Entitlement IT (M).
289
0.167* M 0.311
X Y
-0.480
*p<0.05
Figure C77. Predicting Lives Alone (X) from Level of Injury (Y) through Dangerous World
IT (M).
-0.028 M -0.695
X Y
-0.450
Figure C78. Predicting Lives Alone (X) from Level of Injury (Y) through Women as Sex
Objects IT (M).
-0.090 M 0.007
X Y
-0.421
Figure C79. Predicting Lives Alone (X) from Level of Injury (Y) through Male Sex Drive
-0.008 M -17.876
X Y
-0.446
Figure C80. Predicting Lives Alone (X) from Level of Injury (Y) through Entitlement IT (M).
290
-0.192 M -0.592
X Y
-0.035
Figure C81. Predicting Lives Alone (X) from Offence Outcome (Y) through Dangerous
World IT (M).
0.074 0.079
M
X Y
0.041
Figure C82. Predicting Lives Alone (X) from Offence Outcome (Y) through Women as Sex
Objects IT (M).
0.023 0.862
M
X Y
0.026
Figure C83. Predicting Lives Alone (X) from Offence Outcome (Y) through Male Sex Drive
0.008 M -17.775
X Y
0.058
Figure C84. Predicting Lives Alone (X) from Offence Outcome (Y) through Entitlement IT
(M).
291
-0.001 M -0.588
X Y
-0.032
Figure C85. Predicting Lives Alone (X) from Victim Age (Y) through Dangerous World IT
(M).
0.002 0.106
M
X Y
-0.032
Figure C86. Predicting Lives Alone (X) from Victim Age (Y) through Women as Sex
Objects IT (M).
-0.001 M 0.860
X Y
-0.033
Figure C87. Predicting Lives Alone (X) from Victim Age (Y) through Male Sex Drive
-0.001 M -17.816
X Y
-0.032
Figure C88. Predicting Lives Alone (X) from Victim Age (Y) through Entitlement IT (M).
292
Lives Alone from Weapon Use
0.311* M -0.630
X Y
0.382
*p=0.05
Figure C89. Predicting Lives Alone (X) from Weapon Being Used (Y) through Dangerous
World IT.
-0.082 0.103
M
X Y
0.249
Figure C90. Predicting Lives Alone (X) from Weapon Being Used (Y) through Women as
Sex Objects IT (M).
-0.123 M 0.903
X Y
0.373
Figure C91. Predicting Lives Alone (X) from Weapon Being Used (Y) through Male Sex
0.052 M -17.821
X Y
0.317
Figure C92. Predicting Lives Alone (X) from Weapon Being Used (Y) through Entitlement
IT (M).
293
0.064 M -0.503
X Y
-0.452
Figure C93. Predicting Lives Alone (X) from Sexual Penetration (Y) through Dangerous
World IT (M).
-0.027 M 0.043
X Y
-0.485
Figure C94. Predicting Lives Alone (X) from Sexual Penetration (Y) through Women as Sex
Objects IT (M).
-0.022 M 0.836
X Y
-0.461
Figure C95. Predicting Lives Alone (X) from Sexual Penetration (Y) through Male Sex Drive
-0.030 -17.948
M
X Y
-0.529
Figure C96. Predicting Lives Alone (X) from Sexual Penetration (Y) through Entitlement IT
(M).
294
Table C22. Indirect Effects of Level of Aggression on Lives Alone through Implicit
Bootstrapping

CI
Dangerous World -0.052 0.144 -0.297 0.064 -0.385 0.037 -0.366 0.041
Women as Sex
0.000 0.150 -0.118 0.130 -0.104 0.157 -0.127 0.124
Objects
Male Sex Drive
Uncontrollable -0.013 0.066 -0.163 0.107 -0.208 0.077 -0.211 0.074
Entitlement 0.478 0.658 -0.613 1.919 -0.579 2.028 -0.484 2.313
Table C23. Indirect Effects of Level of Injury on Lives Alone through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World 0.052 0.248 -0.181 0.393 -0.181 0.395 -0.223 0.349
Women as Sex
0.020 0.494 -1.011 1.385 -0.546 1.942 -1.099 1.336
Objects
Male Sex Drive
Uncontrollable -0.001 0.299 -0.204 0.333 -0.229 0.291 -0.250 0.261
Entitlement 0.143 0.581 -0.903 1.403 -0886 1.478 -0.807 1.609
Table C24. Indirect Effects of Offence Outcome on Lives Alone through Implicit Variables
(ab paths)
Bootstrapping

CI
Dangerous World 0.114 0.288 -0.087 0.718 -0.047 1.125 -0.051 1.012
Women as Sex
0.006 0.269 -0.228 0.181 -0.154 0.228 -0.135 0.266
Objects
Male Sex Drive
Uncontrollable 0.019 0.133 -0.202 0.241 -0.152 0.318 -0.158 0.311
Entitlement -0.139 0.942 -1.903 1.778 -1.782 1.902 -1.693 2.007
295
Table C25. Indirect Effects of Victim Age on Lives Alone through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World 0.001 0.005 -0.005 0.008 -0.004 0.010 -0.004 0.011
Women as Sex
0.000 0.009 -0.006 0.006 -0.004 0.008 -0.003 0.008
Objects
Male Sex Drive
Uncontrollable -0.001 0.003 -0.008 0.004 -0.009 0.002 -0.009 0.003
Entitlement 0.009 0.017 -0.025 0.043 -0.025 0.043 -0.024 0.044
Table C26. Indirect Effects of Weapon Used on Lives Alone through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World -0.196 0.411 -1.025 0.101 -1.351 0.062 -1.337 0.062
Women as Sex
-0.008 0.332 -0.188 0.275 -0.297 0.144 -0.381 0.123
Objects
Male Sex Drive
Uncontrollable -0.111 0.153 -0.448 0.127 -0.561 0.075 -0.520 0.087
Entitlement -0.925 1.395 -4.033 1.368 -4.355 1.222 -5.200 1.032
Table C27. Indirect Effects of Sexual Penetration on Lives Alone through Implicit Variables
(ab paths)
Bootstrapping

CI
Dangerous World -0.032 0.088 -0.202 0.051 -0.278 0.026 -0.254 0.030
Women as Sex
-0.001 0.090 -0.065 0.089 -0.083 0.071 -0.094 0.061
Objects
Male Sex Drive
Uncontrollable -0.018 0.073 -0.148 0.081 -0.183 0.059 -0.160 0.069
Entitlement 0.539 0.235 0.166 1.080 0.178 1.131 0.202 1.219
296
Table C28. Total Effects of Offence Variables on Lives Alone (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.047 0.560 0.083 0.934 0.007
Victim Age -0.032 0.020 -1.603 0.109 2.571
Weapon Used 0.240 0.640 0.376 0.707 0.141
297
Employed from Level of Aggression Used
0.106 M 0.109
X Y
-0.214
Regression Summary: -2LL=126139, Model LL=0.704, McFadden=0.006, CoxSnell=0.008, Nagelkrk=0.010
Figure C97. Predicting Employed (X) from Aggression (Y) through Dangerous World IT
(M).
-0.038 M -0.160
X Y
-0.209
Figure C98. Predicting Employed (X) from Aggression (Y) through Women as Sex Objects
IT (M).
-0.015 M 0.576
X Y
-0.197
Figure C99. Predicting Employed (X) from Aggression (Y) through Male Sex Drive
-0.027 M -1.241
X Y
-0.243
Figure C100. Predicting Employed (X) from Aggression (Y) through Entitlement IT (M).
298
Employed from Level of Injury
0.167* M 0.244
X Y
0.067
* p<0.05
Figure C101. Predicting Employed (X) from Level of Injury (Y) through Dangerous World
IT (M).
-0.028 -0.482
M
X Y
0.092
Figure C102. Predicting Employed (X) from Level of Injury (Y) through Women as Sex
Objects IT (M).
-0.090 0.106
M
X Y
0.138
Figure C103. Predicting Employed (X) from Level of Injury (Y) through Male Sex Drive
Uncontrollable IT (M.
-0.008 M -0.454
X Y
0.102
Figure C104. Predicting Employed (X) from Level of Injury (Y) through Entitlement IT (M).
299
Employed from Offence Outcome
-0.192 M 0.076
X Y
0.063
Figure C105. Predicting Employed (X) from Offence Outcome (Y) through Dangerous World
IT (M).
0.074 M -0.135
X Y
0.059
Figure C106. Predicting Employed (X) from Offence Outcome (Y) through Women as Sex
Objects IT (M).
0.023 0.581
M
X Y
0.037
Figure C107. Predicting Employed (X) from Offence Outcome (Y) through Male Sex Drive
0.008 M -1.158
X Y
0.059
Figure C108. Predicting Employed (X) from Offence Outcome (Y) through Entitlement IT
(M).
300
Employed from Victim Age
-0.001 M 0.073
X Y
0.006
Figure C109. Predicting Employed (X) from Victim Age (Y) through Dangerous World IT
(M).
0.002 -0.147
M
X Y
0.006
Regression Summary: -2LL=126.502 Model LL=0.342, McFadden=0.003, CoxSnell=0.004, Nagelkrk=0.005
Figure C110. Predicting Employed (X) from Victim Age (Y) through Women as Sex Objects
IT (M).
-0.001 0.599
M
X Y
0.007
Figure C111. Predicting Employed (X) from Victim Age (Y) through Male Sex Drive
-0.001 M -1.142
X Y
0.005
Figure C112. Predicting Employed (X) from Victim Age (Y) through Entitlement IT (M).
301
Employed from Weapon Use
0.311* M 0.084
X Y
-0.113
*p<0.05
Figure C113. Predicting Employed (X) from Weapon Being Used (Y) through Dangerous
World IT (M).
-0.082 -0.087
M
X Y
-0.098
Figure C114. Predicting Employed (X) from Weapon Being Used (Y) through Women as Sex
Objects IT (M).
-0.123 M 0.581
X Y
-0.019
Figure C115. Predicting Employed (X) from Weapon Being Used (Y) through Male Sex
0.052 M -1.152
X Y
-0.028
Figure C116. Predicting Employed (X) from Weapon Being Used (Y) through Entitlement IT
(M).
302
Employed from Sexual Penetration
0.064 M 0.086
X Y
-0.098
Figure C117. Predicting Employed (X) from Sexual Penetration (Y) through Dangerous
World IT (M).
-0.027 -0.144
M
X Y
-0.097
Figure C118. Predicting Employed (X) from Sexual Penetration (Y) through Women as Sex
Objects IT (M).
-0.022 M 0.575
X Y
-0.082
Figure C119. Predicting Employed (X) from Sexual Penetration (Y) through Male Sex Drive
-0.030 M -1.220
X Y
-0.130
Figure C120. Predicting Employed (X) from Sexual Penetration (Y) through Entitlement IT
(M).
303
Table C29. Indirect Effects of Level of Aggression on Employed through Implicit Variables
(ab paths)
Bootstrapping

CI
Dangerous World 0.012 0.048 -0.080 0.125 -0.063 0.143 -0.059 0.149
Women as Sex
0.006 0.049 -0.069 0.095 -0.040 0.155 -0.041 0.147
Objects
Male Sex Drive
Uncontrollable -0.009 0.053 -0.124 0.083 -0.165 0.057 -0.167 0.056
Entitlement 0.033 0.304 -0.118 1.095 -0.053 1.347 -0.055 1.314
Table C30. Indirect Effects of Level of Injury on Employed through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World 0.041 0.092 -0.148 0.229 -0.128 0.258 -0.120 0.275
Women as Sex
0.014 0.151 -0.108 0.178 -0.048 0.815 -0.054 0.557
Objects
Male Sex Drive
Uncontrollable -0.032 0.072 -0.195 0.098 -0.251 0.067 -0.238 0.073
Entitlement 0.004 0.239 -0.402 0.741 -0.275 0.895 -0.481 0.658
Table C31. Indirect Effects of Offence Outcome on Employed through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World -0.015 0.089 -0.244 0.126 -0.282 0.111 -0.287 0.107
Women as Sex
-0.010 0.061 -0.129 0.129 -0.199 0.076 -0.195 0.078
Objects
Male Sex Drive
Uncontrollable 0.013 0.080 -0.186 0.156 -0.108 0.220 -0.112 0.213
Entitlement -0.009 0.356 -1.234 0.369 -1.196 0.409 -0.833 0.794
304
Table C32. Indirect Effects of Victim Age on Employed through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World -0.000 0.001 -0.004 0.002 -0.003 0.002 -0.003 0.003
Women as Sex
-0.000 0.002 -0.004 0.003 -0.007 0.002 -0.007 0.002
Objects
Male Sex Drive
Uncontrollable -0.001 0.002 -0.006 0.003 -0.007 0.002 -0.007 0.002
Entitlement 0.001 0.007 -0.006 0.025 -0.003 0.030 -0.003 0.031
Table C33. Indirect Effects of Weapon Used on Employed through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World 0.026 0.142 -0.190 0.396 -0.158 0.434 -0.156 0.470
Women as Sex
0.011 0.071 -0.159 0.150 -0.085 0.229 -0.086 0.223
Objects
Male Sex Drive
Uncontrollable -0.071 0.108 -0.343 0.099 -0.423 0.053 -0.401 0.062
Entitlement -0.060 0.472 -1.300 0.824 -2.497 0.155 -2.645 0.131
Table 34. Indirect Effects of Sexual Penetration on Employed through Implicit Variables (ab
paths)
Bootstrapping

CI
Dangerous World 0.006 0.037 -0.081 0.079 -0.046 0.126 -0.038 0.146
Women as Sex
0.004 0.025 -0.060 0.050 -0.032 0.075 -0.033 0.075
Objects
Male Sex Drive
Uncontrollable -0.013 0.043 -0.124 0.060 -0.150 0.042 -0.134 0.049
Entitlement 0.037 0.148 -0.021 0.564 -0.025 0.533 -0.019 0.575
305
Table C35. Total Effects of Offence Variables on Employed (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.049 0.437 0.112 0.911 0.013
Victim Age 0.006 0.011 0.504 0.615 0.254
Weapon Used -0.087 0.521 -0.167 0.867 0.028
306
Appendix G: Content Dictionary for Motivations Coding
1) Angry
 Women are the central focus of the offender’s anger
 Acts within the offence are meant to punish and physically harm the victim
 Often the offender will use excessive violence that is beyond that necessary to
control the victim
 These attacks are often preceded or induced by life circumstances
 Not usually planned
 Often a stranger victim or prostitute; victim who is immediately available
 Victim is often seriously injured
2) Sexually Opportunistic
 Impulsive predatory act
 Influenced by contextual and situational factors (e.g., rape of a woman during a
burglary; woman who they met at a bar)
 Often used alcohol before offending
 Minimal force used throughout the offence
 Victim typically an acquaintance
3) Sexually Compensatory
 Aim of assault is to control woman; to compensate for underlying feelings of
inadequacy
 Overwhelming rape fantasies preceding offence
 Offences are planned
 Instrumental aggression (only what’s necessary to control victim and complete
offence)
4) Sadistic
 Fusion of sex and aggression
 Engage in physically damaging and degrading behaviours towards the victim
(e.g., use of restraints, blindfolding, beating, burning, strangulation, torture,
mutilation)
 Have and engage in elaborate deviant sexual fantasies
 Offences are planned and victims tend to be selected
 Prolonged contact with victim
307
Appendix H: Figures and Tables of the Motivations Mediation Analyses
-138.987 M 0.006
X Y
0.249
Figure F1. Predicting Perpetrator age (X) from Aggression (Y) through Angry Motivation
(M).
-139.076 0.006
M
X Y
0.251
Figure F2. Predicting Perpetrator age (X) from Aggression (Y) through Sadistic Motivation
(M). -139.154 0.006
M
X Y
-0.662
Figure F3. Predicting Perpetrator age (X) from Aggression (Y) through Sexually
Compensatory Motivation (M).
-139.248 M 0.006
X Y
0.251
Figure F4. Predicting Perpetrator age (X) from Aggression (Y) through Sexually
Opportunistic Motivation (M).
308
0.049 M -0.283
X Y
-2.050
Figure F5. Predicting Perpetrator age (X) from Level of Injury (Y) through Angry Motivation
(M).
0.038 M 0.824
X Y
-2.095*
* p<0.05
Figure F6. Predicting Perpetrator age (X) from Level of Injury (Y) through Sadistic
Motivation (M).
-0.034 1.779
M
X Y
-2.004
Figure F7. Predicting Perpetrator age (X) from Level of Injury (Y) through Sexually
-0.053 -1.358
M
X Y
-2.135*
* p<0.05
Figure F8. Predicting Perpetrator age (X) from Level of Injury (Y) through Sexually
309
-3.371 M 0.006
X Y
5.499*
* p<0.01
Figure F9. Predicting Perpetrator age (X) from Offence Outcome (Y) through Angry
Motivation (M).
-3.301 0.006
M
X Y
5.499*
* p<0.05
Figure F10. Predicting Perpetrator age from Offence Outcome through Sadistic Motivation.
-3.216 M 0.006
X Y
5.498*
* p<0.01
Figure F11. Predicting Perpetrator age (X) from Offence Outcome (Y) through Sexually
-3.162 0.006
M
X Y
5.498*
* p<0.01
Figure F12. Predicting Perpetrator age (X) from Offence Outcome (Y) through Sexually
310
0.189 0.006
M
X Y
-0.041
Figure F13. Predicting Perpetrator age (X) from Victim Age (Y) through Angry Motivation
(M).
0.195 0.006
M
X Y
-0.041
Figure F14. Predicting Perpetrator age (X) from Victim Age (Y) through Sadistic Motivation
(M).
0.191 0.006
M
X Y
-0.041
Figure F15. Predicting Perpetrator age (X) from Victim Age (Y) through Sexually
0.191 0.006
M
X Y
-0.041
Figure F16. Predicting Perpetrator age (X) from Victim Age (Y) through Sexually
311
-73.896 0.005
M
X Y
-2.378
Figure F17. Predicting Perpetrator age (X) from Weapon Being Used (Y) through Angry
Motivation (M).
-73.829 M 0.005
X Y
-2.845
Figure F18. Predicting Perpetrator age (X) from Weapon Being Used (Y) through Sadistic
Motivation (M).
-74.104 0.005
M
X Y
-2.377
Figure F19. Predicting Perpetrator age (X) from Weapon Being Used (Y) through Sexually
-74.097 0.005
M
X Y
-2.377
Figure F20. Predicting Perpetrator age (X) from Weapon Being Used (Y) through Sexually
312
-27.161 0.006
M
X Y
0.347
Figure F21. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Angry
Motivation (M).
-27.235 0.006
M
X Y
0.348
*p<0.06
Figure F22. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Sadistic
Motivation (M).
-27.329 0.006
M
X Y
0.349
Figure F23. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Sexually
-27.344 0.006
M
X Y
0.349
Figure F24. Predicting Perpetrator age (X) from Sexual Penetration (Y) through Sexually
313
Table F1. Indirect Effects of Level of Aggression on Perpetrator Age through Motive
Bootstrapping
Product of
Angry -0.874 0.718 -1.218 -3.591 1.161 -3.927 1.006 -4.946 0.776
Sadistic -0.876 0.718 -1.220 -3.610 1.167 -4.027 0.981 -4.945 0.756
Sexually Compensatory -0.877 0.719 -1.220 -3.616 1.161 -4.017 0.988 -5.097 0.766
Sexually Opportunistic -0.877 0.719 -1.219 -3.689 1.164 -4.158 0.994 -5.013 0.764
Table F2. Indirect Effects of Level of Injuries on Perpetrator Age through Motive Variables
(ab paths)
Bootstrapping
Product of
Angry -0.014 0.108 -0.128 -0.458 0.351 -0.589 0.272 -0.627 0.257
Sadistic 0.031 0.109 0.285 -0.359 0.748 -0.282 0.881 -0.304 0.833
Sexually Opportunistic 0.072 0.161 0.443 -0.275 0.741 -0.188 0.956 -0.194 0.920
314
Table F3. Indirect Effects of Offence Outcome on Perpetrator Age through Motive Variables
(ab paths)
Bootstrapping
Product of
Angry -0.020 0.334 -0.060 -0.295 1.813 -0.790 0.394 -0.790 0.341
Sadistic -0.020 0.334 -0.059 -0.312 1.950 -0.894 0.405 -0.984 0.339
Sexually Compensatory -0.019 -0.334 -0.058 -0.299 1.890 -0.833 0.440 -0.843 0.371
Sexually Opportunistic -0.019 0.334 -0.057 -0.305 1.821 -1.368 0.402 -1.439 0.343
Table F4. Indirect Effects of Victim Age on Perpetrator Age through Motive Variables (ab
paths)
Bootstrapping
Product of
Angry 0.001 0.009 0.128 -0.041 0.017 -0.016 0.039 -0.012 0.044
Sadistic 0.001 0.009 0.132 -0.043 0.017 -0.016 0.052 -0.012 0.064
Sexually Compensatory 0.001 0.009 0.129 -0.043 0.017 -0.015 0.049 -0.012 0.068
Sexually Opportunistic 0.001 0.009 0.129 -0.043 0.017 -0.016 -0.046 -0.012 -0.123
315
Table F5. Indirect Effects of a Weapon Being Used on Perpetrator Age Motive Implicit
Bootstrapping
Product of
Angry -0.393 0.477 -0.023 -1.900 0.624 -2.129 0.537 -2.717 0.405
Sadistic -0.393 0.477 -0.023 -1.796 0.651 -2.020 0.546 -2.654 0.421
Sexually Compensatory -0.394 0.478 -0.824 -1.870 0.620 -2.059 -0.523 -2.588 -0.414
Sexually Opportunistic -0.394 -0.478 -0.824 -1.795 0.663 -2.000 0.558 -2.521 0.428
Table F6. Indirect Effects of Sexual Penetration on Perpetrator Age through Motive Variables
(ab paths)
Bootstrapping
Product of
Angry -0.165 0.200 -0.826 -0.692 0.214 -0.760 0.188 -1.005 0.145
Sadistic -0.166 0.200 -0.828 -0.709 0.212 -0.791 0.177 -1.002 0.126
Sexually Opportunistic -0.167 0.201 -0.830 -0.725 -0.200 -0.797 0.175 -1.042 0.126
316
Table F7. Total Effects of Offence Variables on Perpetrator Age (c path)
Product of
Coefficients
B SE t p
Offence Outcome 5.479 1.963 2.791 0.007
Victim Age -0.040 0.054 -0.731 0.467
Weapon Used -2.771 2.444 -1.134 0.261
317
Relationship Status from Level of Aggression Used
0.140* 0.504
M
X Y
0.000
*p<0.05
Figure F25. Predicting Relationship Status (X) from Aggression (Y) through Angry
Motivation (M).
0.023 059
M
X Y
0.067
Figure F26. Predicting Relationship Status from Aggression through Sadistic Motivation.
-0.012 -0.617
M
X Y
0.061
Figure F27. Predicting Relationship Status (X) from Aggression (Y) through Sexually
-0.151* -0.222
M
X Y
0.034

*p<0.01
Figure F28. Predicting Relationship Status (X) from Aggression (Y) through Sexually
318
0.032 0.158
M
X Y
-0.272
Figure F29. Predicting Relationship Status (X) from Level of Injury(Y) through Angry
Motivation (M).
0.049 0.084
M
X Y
-0.272
Figure F30. Predicting Relationship Status (X) from Level of Injury (Y) through Sadistic
Motivation (M).
-0.079* -1.515
M
X Y
-0.415
*p <0.05
Figure F31. Predicting Relationship Status (X) from Level of Injury (Y) through Sexually
-0.002 0.790
M
X Y
-0.272
Figure F32. Predicting Relationship Status (X) from Level of Injury(Y) through Sexually
319
-0.066 0.593
M
X Y
0.943*
*p <0.05
Figure F33. Predicting Relationship Status (X) from Offence Outcome(Y) through Angry
Motivation (M).
-0.050 0.130
M
X Y
0.895
Figure F34. Predicting Relationship Status (X) from Offence Outcome (Y) through Sadistic
Motivation (M).
0.079 -0.790
M
X Y
0.972*
*p <0.05
Figure F35. Predicting Relationship Status (X) from Offence Outcome (Y) through Sexually
0.038 -0.305
M
X Y
0.903*
* p <0.05
Figure F36. Predicting Relationship Status (X) from Offence Outcome (Y) through Sexually
320
-0.002 0.465
M
X Y
-0.013
Figure F37. Predicting Relationship Status (X) from Victim Age(Y) through Angry
Motivation (M).
0.003 0.137
M
X Y
-0.015
Figure F38. Predicting Relationship Status (X) (Y) from Victim Age through Sadistic
Motivation (M).
-0.002 -0.707
M
X Y
-0.016
Figure F39. Predicting Relationship Status (X) from Victim Age(Y) through Sexually
0.002 -0.194
M
X Y
-0.013
Figure F40. Predicting Relationship Status (X) from Victim Age(Y) through Sexually
321
Relationship Status from Weapon Use
0.093 0.540
M
X Y
-0.474
Figure F41. Predicting Relationship Status (X) from Weapon Being Used (Y) through Angry
Motivation (M).
0.171 0.138
M
X Y
-0.441
Figure F42. Predicting Relationship Status (X) from Weapon Being Used (Y) through
Sadistic Motivation (M).
-0.145 -0.721
M
X Y
-0.533
Sexually Compensatory Motivation (M).
-0.120 -0.302
M
X Y
-0.456
Sexually Opportunistic Motivation (M).
322
0.070 M 0.595
X Y
-0.264
Figure F45. Predicting Relationship Status (X) from Sexual Penetration (Y) through Angry
Motivation (M).
0.059 M 0.143
X Y
-0.226
Figure F46. Predicting Relationship Status (X) from Sexual Penetration (Y) through Sadistic
Motivation (M).
-0.057 M -0.738
X Y
-0.264
Figure F47. Predicting Relationship Status (X) from Sexual Penetration (Y) through Sexually
-0.072 M -0.358
X Y
-0.245
Figure F48. Predicting Relationship Status (X) from Sexual Penetration (Y) through Sexually
323
Table F8. Indirect Effects of Level of Aggression on Relationship Status through Motive
Bootstrapping

CI
Angry 0.070 0.081 -0.076 0.253 -0.042 0.300 -0.045 0.293
Sadistic 0.001 0.036 -0.084 0.076 -0.059 0.107 -0.068 0.091
Sexually Compensatory 0.008 0.052 -0.073 0.118 -0.048 0.170 -0.043 0.189
Sexually Opportunistic 0.033 0.121 -0.175 0.257 -0.148 0.290 -0.145 0.294
Table F9. Indirect Effects of Level of Injury on Relationship Status through Motive Variables
(ab paths)
Bootstrapping

CI
Angry 0.005 0.041 -0.080 0.099 -0.050 0.146 -0.055 0.132
Sadistic 0.004 0.123 -0.088 0.156 -0.086 0.161 -0.113 0.131
324
Table F10. Indirect Effects of Offence Outcome on Relationship Status through Motive
Bootstrapping

CI
Angry -0.039 0.093 -0.258 0.129 -0.368 0.066 -0.360 0.069
Sadistic -0.007 0.092 -0.192 0.106 -0.254 0.084 -0.212 0.098
Sexually Compensatory -0.062 0.117 -0.224 0.130 -0.325 0.040 -0.337 0.035
Table F11. Indirect Effects of Victim Age on Relationship Status through Motive Variables
(ab paths)
Bootstrapping

CI
Angry -0.001 0.002 -0.007 0.003 -0.010 0.001 -0.009 0.001
Sadistic 0.003 0.003 -0.003 0.007 -0.003 0.009 -0.003 0.008
325
Table F12. Indirect Effects of Weapon Used on Relationship Status through Motive Variables
(ab paths)
Bootstrapping

CI
Angry 0.051 0.144 -0.136 0.336 -0.072 0.449 -0.075 0.432
Sadistic 0.024 0.148 -0.206 0.316 -0.157 0.406 -0.171 0.375
Table F13. Indirect Effects of Sexual Penetration on Relationship Status through Motive
Bootstrapping

CI
Lower Upper Lower Upper Lower Uppe
r
Angry 0.042 0.059 -0.039 -0.188 -0.024 0.233 -0.027 0.225
Sadistic 0.008 0.057 -0.085 0.112 -0.055 0.156 -0.059 0.142
326
Table F14. Total Effects of Offence Variables on Relationship Status (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.888 0.457 1.941 0.052 3.769
Victim Age -0.014 0.012 -1.186 0.236 1.406
Weapon Used -0.417 0.529 -0.788 0.431 0.621
327
Previous Convictions from Level of Aggression Used
0.140 M 0.021
X Y
0.375
*p<0.05
Figure F49. Predicting Previous Convictions (X) from Aggression (Y) through Angry
Motivation (M).
0.023 0.140
M
X Y
0.376
Figure F50. Predicting Previous Convictions (X) from Aggression (Y) through Sadistic
Motivation (M).
-0.012 -1.161
M
X Y
0.385
Figure 51. Predicting Previous Convictions (X) from Aggression (Y) through Sexually
-0.151* M 1.316
X Y
0.575
*p<0.01
Figure F52. Predicting Previous Convictions (X) from Aggression (Y) through Sexually
328
Previous Convictions from Level of Injury
0.032 M 0.169
X Y
0.541*
*p<0.05
Figure F53. Predicting Previous Convictions (X) from Level of Injury (Y) through Angry
Motivation (M).
0.049 M -0.322
X Y
0.566*
* p<0.05
Figure F54. Predicting Previous Convictions (X) from Level of Injury (Y) through Sadistic
Motivation (M).
-0.079* -1.134
M
X Y
0.459
*p<0.05
Figure F55. Predicting Previous Convictions (X) from Level of Injury (Y) through Sexually
-0.002 M 1.367
X Y
0.576*
*p<0.05
Figure F56. Predicting Previous Convictions (X) from Level of Injury (Y) through Sexually
329
Previous Convictions from Offence Outcome
-0.066 M 0.166
X Y
-0.040
Figure F57. Predicting Previous Convictions (X) from Offence Outcome (Y) through Angry
Motivation (M).
-0.050 0.163
M
X Y
-0.043
Figure F58. Predicting Previous Convictions (X) from Offence Outcome (Y) through Sadistic
Motivation (M).
0.079 -1.165
M
X Y
-0.061
Figure F59. Predicting Previous Convictions (X) from Offence Outcome (Y) through
0.038 M 0.910
X Y
-0.080
Figure F60. Predicting Previous Convictions (X) from Offence Outcome (Y) through
330
Previous Convictions from Victim Age
-0.002 M 0.234
X Y
0.053*
*p<0.05
Figure F61. Predicting Previous Convictions from Victim Age through Angry Motivation.
-0.003 M 0.026
X Y
0.054*
*p<0.05
Figure F62. Predicting Previous Convictions from Victim Age through Sadistic Motivation.
-0.002 M -1.060
X Y
0.051*
*p<0.05
Figure 63. Predicting Previous Convictions from Victim Age through Sexually
Compensatory Motivation.
0.002 M 0.881
X Y
0.054*
*p<0.05
Figure F64. Predicting Previous Convictions from Victim Age through Sexually
Opportunistic Motivation.
331
Previous Convictions from Weapon Use
0.093 M 0.169
X Y
-0.011
Figure F65. Predicting Previous Convictions from Weapon Being Used through Angry
Motivation.
0.171 0.170
M
X Y
-0.024
Figure F66. Predicting Previous Convictions from Weapon Being Used through Sadistic
Motivation.
-0.145 M -1.196
X Y
-0.213
Figure F67. Predicting Previous Convictions from Weapon Being Used through Sexually
-0.120 M 0.917
X Y
0.095
Figure F68. Predicting Previous Convictions from Weapon Being Used through Sexually
332
Previous Convictions from Sexual Penetration
0.070 M 0.263
X Y
-0.271
Figure F69. Predicting Previous Convictions from Sexual Penetration through Angry
Motivation.
0.059 M 0.266
X Y
-0.267
Figure F70. Predicting Previous Convictions from Sexual Penetration through Sadistic
Motivation.
-0.057 M -1.348*
X Y
-0.360
Figure F71. Predicting Previous Convictions from Sexual Penetration through Sexually
-0.072 M 0.806
X Y
-0.204
Figure F72. Predicting Previous Convictions from Sexual Penetration through Sexually
333
Table F15. Indirect Effects of Level of Aggression on Previous Convictions through
Motivational Variables (ab paths)
Bootstrapping

CI
Angry 0.003 0.119 -0.180 0.216 -0.178 0.218 -0.183 0.214
Sadistic 0.003 0.158 -0.108 0.125 -0.064 0.382 -0.091 0.146
Table F16. Indirect Effects of Level of Injury on Previous Convictions through Motivational
Bootstrapping

CI
Angry 0.005 0.061 -0.105 0.116 -0.068 0.153 -0.082 0.147
Sadistic -0.016 0.182 -0.290 0.193 -0.285 0.248 -0.253 0.387
Table F17. Indirect Effects of Offence Outcome on Previous Convictions through

Bootstrapping

CI
Angry -0.011 0.086 -0.221 0.139 -0.303 0.087 -0.284 0.095
Sadistic -0.008 0.210 -0.170 0.194 -0.940 0.082 -0.354 0.097
334
Table F18. Indirect Effects of Victim Age on Previous Convictions through Motivational
Bootstrapping

CI
Angry -0.001 0.002 -0.006 0.003 -0.008 0.002 -0.008 0.002
Sadistic 0.000 0.011 -0.005 0.009 -0.005 0.011 -0.005 0.008
Table F19. Indirect Effects of Weapon Used on Previous Convictions through Motivational
Bootstrapping

CI
Angry 0.016 0.100 -0.171 0.259 -0.099 0.383 -0.106 0.367
Sadistic 0.029 0.379 -0.266 0.476 -0.176 1.037 -0.207 0.606
Table F20. Indirect Effects of Sexual Penetration on Previous Convictions through

Bootstrapping

CI
Angry 0.018 0.59 -0.072 0.158 -0.08 0.216 -0.054 0.203
Sadistic 0.016 0.161 -0.082 0.198 -0.049 0.514 -0.059 0.290
Sexually Compensatory 0.076 0.060 -0.008 0.185 0.001 0.200 0.003 0.204
335
Table F21. Total Effects of Offence Variables on Previous Convictions (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome -0.051 0.535 -0.095 0.925 0.009
Victim Age 0.054 0.025 2.141 0.033 4.582
Weapon Used 0.005 0.633 0.007 0.994 0.000
336
Lives Alone from Level of Aggression Used
0.140* M 0.129
X Y
-0.596*
*p<0.05
Figure F73. Predicting Lives Alone from Aggression through Angry Motivation.
0.023 M -0.302
X Y
-0.574
Figure F74. Predicting Lives Alone from Aggression through Sadistic Motivation.
-0.012 M 0.163
X Y
-0.576
Figure F75. Predicting Lives Alone from Aggression through Sexually Compensatory
Motivation.
-0.151* -0.008
M
X Y
-0.578
*p<0.05
Figure F76. Predicting Lives Alone from Aggression through Sexually Opportunistic
Motivation.
337
0.032 M 0.545
X Y
-0.455
Figure F77. Predicting Lives Alone from Level of Injury through Angry Motivation.
0.049 M -0.746
X Y
-0.394
Figure F78. Predicting Lives Alone from Level of Injury through Sadistic Motivation.
-0.079* M -0.854
X Y
-0.4
*p<0.05
Figure F79. Predicting Lives Alone from Level of Injury through Sexually Compensatory
Motivation.
-0.002 0.322
M
X Y
-0.420
Figure F80. Predicting Lives Alone from Level of Injury through Sexually Opportunistic
Motivation.
338
-0.0 M -0.127
X Y
0.006
Figure F81. Predicting Lives Alone from Offence Outcome through Angry Motivation.
-0.050 M -0.331
X Y
-0.001
Figure F82. Predicting Lives Alone from Offence Outcome through Sadistic Motivation.
0.079 M 0.176
X Y
0.000
Regression Summary: -2LL=84.182, Model LL=059, McFadden=0.001, CoxSnell=0.001, Nagelkrk=0.001
Figure F83. Predicting Lives Alone from Offence Outcome through Sexually Compensatory
Motivation.
0.038 M 0.357
X Y
0.000
Figure F84. Predicting Lives Alone from Offence Outcome through Sexually Opportunistic
Motivation.
339
-0.002 M -0.181
X Y
-0.029
Figure F85. Predicting Lives Alone from Victim Age through Angry Motivation.
0.003 M -0.242
X Y
-0.028
Figure F86. Predicting Lives Alone from Victim Age through Sadistic Motivation.
-0.002 M -0.070
X Y
-0.028
Figure F87. Predicting Lives Alone from Victim Age through Sexually Compensatory
Motivation.
0.002 M 0.437
X Y
-0.029
Figure F88. Predicting Lives Alone from Victim Age through Sexually Compensatory
Motivation.
340
Lives Alone from Weapon Use
0.093 M -0.148
X Y
0.285
Figure F89. Predicting Lives Alone from Weapon Being Used through Angry Motivation.
0.171 -0.389
M
X Y
0.335
Figure F90. Predicting Lives Alone from Weapon Being Used through Sadistic Motivation.
-0.145 0.233
M
X Y
0.307
Figure F91. Predicting Lives Alone from Weapon Being Used through Sexually
-0.120 M -0.401
X Y
0.324
Figure F92. Predicting Lives Alone from Weapon Being Used through Sexually
341
0.070 -0.036
M
X Y
-0.477
Figure F93. Predicting Lives Alone from Sexual Penetration through Angry Motivation.
0.059 M -0.223
X Y
-0.467
Figure F94. Predicting Lives Alone from Sexual Penetration through Sadistic Motivation.
-0.057 0.053
M
X Y
-0.477
Figure F95. Predicting Lives Alone from Sexual Penetration through Sexually Compensatory
Motivation.
-0.072 0.212
M
X Y
-0.463
Figure F96. Predicting Lives Alone from Sexual Penetration through Sexually Opportunistic
Motivation.
342
Table F22. Indirect Effects of Level of Aggression on Lives Alone through Motivational
Bootstrapping

CI
Angry 0.018 0.173 -0.195 0.261 -0.173 0.285 -0.172 0.287
Sadistic 0.007 0.303 -0.834 0.132 -1.898 0.067 -1.076 0.107
Table F23. Indirect Effects of Level of Injury on Lives Alone through Motivational Variables
(ab paths)
Bootstrapping

CI
Angry 0.018 0.167 -0.103 0.219 -0.054 0.479 -0.058 0.384
Sadistic -0.037 0.636 -2.050 0.224 -2.162 0.171 -1.652 0.995
Table F24. Indirect Effects of Offence Outcome on Lives Alone through Motivational
Bootstrapping

CI
Angry 0.008 0.134 -0.158 0.198 -0.103 0.290 -0.114 0.268
Sadistic 0.018 0.420 -0.153 0.724 -0.087 2.566 -0.109 1.714
343
Table F25. Indirect Effects of Victim Age on Lives Alone through Motivational Variables
(ab paths)
Bootstrapping

CI
Angry 0.000 0.005 -0.004 0.006 -0.003 0.008 -0.003 0.008
Sadistic -0.001 0.013 -0.031 0.007 -0.089 0.003 -0.070 0.004
Table F26. Indirect Effects of Weapon Used on Lives Alone through Motivational Variables
(ab paths)
Bootstrapping

CI
Angry -0.014 0.182 -0.274 0.176 -0.413 0.114 -0.375 0.123
Sadistic -0.067 0.662 -2.043 0.219 -3.641 0.149 -2.830 0.184
Table F27. Indirect Effects of Sexual Penetration on Lives Alone through Motivational
Bootstrapping

CI
Angry -0.003 0.091 -0.141 0.105 -0.166 0.094 -0.152 0.099
Sadistic -0.013 0.342 -1.132 0.093 -1.951 0.068 -1.371 0.087
344
Table F28. Total Effects of Offence Variables on Lives Alone (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome 0.0144 0.569 0.025 0.979 0.001
Victim Age -0.028 0.020 -1.397 0.162 1.952
Weapon Used 0.271 0.646 0.420 0.675 0.176
345
Employed from Level of Aggression Used
0.140* M 0.056
X Y
-0.0451
Regression Summary: -2LL=122.539, Model LL=0.041 0, McFadden=0.000, CoxSnell=0.001, Nagelkrk=0.001
*p < 0.05
Figure F97. Predicting Employed from Aggression through Angry Motivation.
0.023 M 0.227
X Y
-0.049
Figure F98. Predicting Employed from Aggression through Sadistic Motivation.
-0.012 M 0.450
X Y
-0.038
Figure F99. Predicting Employed from Aggression through Sexually Compensatory

Motivation.
-0.151* M -0.772
X Y
-0.165
*p < 0.05
Figure F100. Predicting Employed from Aggression through Sexually Opportunistic

Motivation.
346
Employed from Level of Injury
0.032 M 0.099
X Y
0.102
* p<0.05
Figure F101. Predicting Employed from Level of Injury through Angry Motivation.
0.049 M 0.109
X Y
0.100
Figure F102. Predicting Employed from Level of Injury through Sadistic Motivation.
-0.079* M -0.065
X Y
0.100
* p<0.05
Figure F103. Predicting Employed from Level of Injury through Sexually Compensatory
Motivation.
-0.002 0.252
M
X Y
0.105
Figure F104. Predicting Employed from Level of Injury through Sexually Opportunistic
Motivation.
347
Employed from Offence Outcome
-0.066 M 0.034
X Y
-0.067
Figure F105. Predicting Employed from Offence Outcome through Angry Motivation.
-0.050 M 0.219
X Y
-0.058
Figure F106. Predicting Employed from Offence Outcome through Sadistic Motivation.
0.079 M 0.466
X Y
-0.105
Figure F107. Predicting Employed from Offence Outcome through Sexually Compensatory
Motivation.
0.038 -0.673
M
X Y
-0.044
Figure F108. Predicting Employed from Offence Outcome through Sexually Opportunistic
Motivation.
348
Employed from Victim Age
-0.002 M 0.048
X Y
0.003
Figure F109. Predicting Employed from Victim Age through Angry Motivation.
0.003 M 0.212
X Y
0.003
Figure F110. Predicting Employed from Victim Age through Sadistic Motivation.
-0.002 M 0.471
X Y
0.004
Figure F111. Predicting Employed from Victim Age through Sexually Compensatory
Motivation.
0.002 -0.691
M
X Y
0.004
Figure F112. Predicting Employed from Victim Age through Sexually Opportunistic
Motivation.
349
Employed from Weapon Use
0.093 M 0.038
X Y
0.003
Figure F113. Predicting Employed from Weapon Being Used through Angry Motivation.
0.171 0.228
M
X Y
-0.033
Figure F114. Predicting Employed from Weapon Being Used through Sadistic Motivation.
-0.145 M 0.464
X Y
0.071
Figure F115. Predicting Employed from Weapon Being Used through Sexually
-0.121 M -0.685
X Y
-0.079
Figure F116. Predicting Employed from Weapon Being Used through Sexually Opportunistic
Motivation.
350
Employed from Sexual Penetration
0.070 0.056
M
X Y
-0.062
Figure F117. Predicting Employed from Sexual Penetration through Angry Motivation.
0.059 0.247
M
X Y
-0.073
Figure F118. Predicting Employed from Sexual Penetration through Sadistic Motivation.
-0.056 M 0.438
X Y
-0.034
Figure F119. Predicting Employed from Sexual Penetration through Sexually Compensatory
Motivation.
-0.072 M -0.727
X Y
-0.113
Figure F120. Predicting Employed from Sexual Penetration through Sexually Opportunistic
Motivation.
351
Table F29. Indirect Effects of Level of Aggression on Employed through Motivational
Bootstrapping

CI
Angry 0.008 0.070 -0.152 0.144 -0.132 0.159 -0.134 0.159
Sadistic 0.005 0.046 -0.068 0.094 -0.044 0.137 -0.052 0.115
Table F30. Indirect Effects of Level of Injury on Employed through Motivational Variables
(ab paths)
Bootstrapping

CI
Angry 0.003 0.040 -0.080 0.096 -0.058 0.125 -0.060 0.121
Sadistic 0.005 0.193 -0.078 0.202 -0.077 0.208 -0.087 0.173
Table F31. Indirect Effects of Offence Outcome on Employed through Motivational

Bootstrapping

CI
Angry -0.002 0.058 -0.127 0.125 -0.157 0.101 -0.152 0.103
Sadistic -0.011 0.079 -0.175 0.091 -0.251 0.059 -0.232 0.065
352
Table F32. Indirect Effects of Victim Age on Employed through Motivational Variables (ab
paths)
Bootstrapping

CI
Angry -0.000 0.002 -0.004 0.003 -0.005 0.002 -0.004 0.003
Sadistic 0.001 0.002 -0.004 0.006 -0.002 0.009 -0.002 0.008
Table F33. Indirect Effects of Weapon Used on Employed through Motivational Variables
(ab paths)
Bootstrapping

CI
Angry 0.004 0.076 -0.159 0.172 -0.130 0.206 -0.134 0.202
Sadistic 0.039 0.128 -0.178 0.360 -0.123 0.415 -0.129 0.432
Table F34. Indirect Effects of Sexual Penetration on Employed through Motivational

Bootstrapping

CI
Angry 0.004 0.042 -0.078 0.100 -0.065 0.117 -0.067 0.116
Sadistic 0.045 0.050 -0.063 0.136 -0.040 0.186 -0.042 0.174
353
Table F35. Total Effects of Offence Variables on Employed (c path)
Product of
Coefficients
B SE z p Wald
Offence Outcome -0.069 0.441 -0.167 0.876 0.025
Victim Age 0.003 0.012 0.261 0.794 0.068
Weapon Used 0.006 0.523 0.002 0.991 0.000
354

Offender Profiling PDF

Uploaded by

Copyright:

Available Formats

Offender Profiling PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Offender Profiling PDF

Uploaded by

Copyright:

Available Formats

OFFENDER PROFILING: A REVIEW, CRITIQUE, AND AN INVESTIGATION OF

THE INFLUENCE OF CONTEXT, PERCEPTION, AND MOTIVATIONS ON

Shannon Lynn Vettor

This unpublished thesis/dissertation is copyright of the author and/or third

Any use made of information contained in this thesis/dissertation must be in

Chapter 1: Offender Profiling: An overview and critique of its assumptions 1

Chapter 2: Directions for Offender Profiling and Current Thesis 34

Chapter 4: Rapists and sexual murderers: Combined pathways to offending 64

Chapter 5, 6, and 7: The Potential Mediators between Crime Scene and 90

Chapter 6: Implicit Theories as a Mediator between Crime Scene and 111

Chapter 8: General Discussion 143

Table Title Page

3.2 Offender Characteristics and their Coding 53

3.3 Offence Characteristics and their Coding 54

3.4 Potential Mediators for Chapters 5, 6, and 7, and their Coding 56

5.2 Indirect Effects of Victim Age on Any Previous Convictions 101

5.3 Indirect Effects of a Weapon Being Used on Perpetrator Age 102

5.5 Total Effects of Offence Variables on Perpetrator Age (c path) 103

5.6 Total Effects of Offence Variables on Relationship Status (c path) 104

5.7 Total Effects of Offence Variables on Any Previous Convictions (c 104

6.2 Indirect Effects of Sexual Penetration on Lives Alone through 121

6.3 Total Effects of Offence Variables on Perpetrator Age (c path) 122

6.4 Total Effects of Offence Variables on Relationship Status (c path) 122

6.5 Total Effects of Offence Variables on Previous Convictions (c path) 123

7.2 Indirect Effects of Sexual Penetration on Previous Convictions 138

7.3 Total Effects of Offence Variables on Perpetrator Age (c path) 139

7.4 Total Effects of Offence Variables on Relationship Status (c path) 139

Figure Title Page

3.2 Total Effect of X on Y 58

1.1.1 Utility of Offender Profiles

1.1.2 Content of Offender Profiles

1.1.3 State of the Literature of Offender Profiling

1.2 The Assumptions of Offender Profiling

1.2.1 The Behavioural Consistency Assumption

1.3 Typographical Approaches

1.3.1 Pragmatic Criminal Investigative Approach

1.3.1.1 Critique of the Early Criminal Investigative Approach

1.3.2 Theory-Led Approach

1) Power Reassurance, represents a planned single rape of the victim with an

1.3.2.1 Critique of the Theory-Led Approach

1.3.3 Clinical Approach

This approach to profiling is heavily reliant on clinical judgment, training, knowledge,

• WHAT (in minute detail) 4. Infer reconstruction

5. Infer Allows importation of factors from

6. Develop psychological 7. Introduce

8. Generate a range 9. Discuss with 10. Produce

1.3.3.1 Critique of the Clinical Approach

1.3.4 Statistical Approach

1.3.4.1 Critique of Statistical Approach

1.4 Critique of Offender Profiling in General

1.4.1 Pragmatic Use and Validity

1.4.2 Quality of Data

In its simplest form Offender Profiling is the determination of offender characteristics

2.1 Rationale for Current Research

Figure 2.1 Revised A  C: Process of Offender Profiling incorporating Perceptions,

Sexual Abuse Static/Historical Stable dynamic risk Deviant sexual

Motivation is not always straightforward or easy to infer from crime scene

A person (A) commits a [rape] if a) he intentionally penetrates the vagina, anus or

3.2 Chapter 4 Methodology