Professional Documents
Culture Documents
Abbie Jean Marono, Sasha Reid, Enzo Yaksic & David Adam Keatley
To cite this article: Abbie Jean Marono, Sasha Reid, Enzo Yaksic & David Adam Keatley (2020):
A Behaviour Sequence Analysis of Serial Killers’ Lives: From Childhood Abuse to Methods of
Murder, Psychiatry, Psychology and Law, DOI: 10.1080/13218719.2019.1695517
The aim of the current research was to provide a new method for mapping the
developmental sequences of serial killers’ life histories. The role of early childhood abuse,
leading to types of serial murder and behaviours involved in the murders, was analysed
using Behaviour Sequence Analysis. A large database (n ¼ 233) of male serial killers with
known childhood abuse (physical, sexual, or psychological) was analysed according to
typologies and crime scene behaviours. Behaviour Sequence Analysis was used to show
significant links between behaviours and events across their lifetime. Sexual, physical, and
psychological abuse often led to distinct crime scene behaviours. The results provide
individual accounts of abuse types and behaviours. The present research highlights the
importance of childhood abuse as a risk factor for serial killers’ behaviours, and provides a
novel and important advance in profiling serial killers and understanding the sequential
progression of their life histories.
Key words: behaviour sequence analysis; crime; homicide; profiling; serial killer.
Homicide is legally defined as the killing of or spree-killing, which is any murder that
another person. Homicide is an all-inclusive occurs at two or more locations with no emo-
term, and there are different subcategories of tional cooling-off period between (Douglas
homicide, such as murder, multicide, and man- et al., 1992).
slaughter. Serial homicide, as defined by the Advances in computational intelligence
Federal Bureau of Investigation, is “the unlaw- and the establishment of large datasets have
ful killing of two or more victims in separate meant that researchers in serial homicide are
events.” Serial homicide is an intentional, pre- moving towards predictive models and under-
meditated act, not a crime carried out on standing. In particular, researchers are now
impulse or in response to a perceived provoca- beginning to develop models to help to predict
tion or threat (Reid, 2016). Athrough it is form who is likely to commit serial homicide, and
of multicide, serial homicide is not to be con- how to interpret offending patterns as a way to
fused with mass murder, which is defined as predict later offending behaviour (Hewitt,
four or more murders that occur in one event, Beauregard, & Martineau, 2016; Ioana, 2013;
with no distinctive time lapse between them, Miller, 2014). Regardless of the type of
Correspondence: David Adam Keatley, School of Law, Murdoch University, Murdoch, WA 6150,
Australia. Email: david.keatley@murdoch.edu.au
School of Psychology, University of Lancaster, UK.
†School of Law, Murdoch University, Murdoch, Australia.
§Department of Sociology, University of Calgary, Canada.
© 2020 The Australian and New Zealand Association of Psychiatry, Psychology and Law
2 A. Marono et al.
prediction, to develop any such model, it is place-specific serial killers. Despite the devel-
important that researchers understand the chain opment of refined typologies, research has
of events that preceded the homicides. One found that there is no such thing as a prototyp-
way in which researchers do this is by group- ical serial killer, consequently limiting the use-
ing related behaviours together, using a fulness of the typologies developed so far
“thematic” or offending style typology (Walters, Drislane, Patrick, & Hickey, 2015).
approach (Grubin et al. 1997). Profiles are created retrospectively – that
is, after a crime had been committed. They are
developed viay a thorough observation of the
Typologies
crime scene, interviews with surviving victims,
Since the 1970s, investigative profilers at the and even wiretappings of taunts made by the
FBI’s Behavioural Science Unit (BSU) have subject to the victims’ families (Douglas et al.,
been analysing crime scenes in the attempt to 1986). However, one limitation to this is that
generate ‘profiles’ of violent offenders. the profiles generated rely, to a large extent,
Profiles consist of aggregated data collected on the use of educated guesses developed on
from several sources, which combine to indi- the basis of data that may be unreliable. While
cate specific characteristics relevant to the profiles are undoubtedly a useful investigative
offender (Douglas, Ressler, Burgess, & tool that should not be overlooked, the accur-
Hartman, 1986). These profiles, in turn, are acy of profiling could still be developed.
meant to aid law enforcement officers in the Studies have suggested that it is important
detection and apprehension of violent to include personal histories and personality
offenders, including serial killers. Originally, factors when proposing an ‘offender profile’
the analysis of crime scenes revealed a dicho- (Hazelwood & Warren, 2000). The FBI’s
tomized classification of a crime that was con- BSU also noted the value of this when they
sidered to be either organized or disorganized conducted a series of extensive interviews
(Hazelwood & Douglas, 1980), the organized with several violent sexual offenders, includ-
typology being a form of murder carried out ing 25 serial killers, in the 1980s (Ressler,
by an individual who appeared to plan the Burgess, & Douglas, 1988). The results from
crime, target the victims specifically, and dis- those interviews have helped to inform the
played control (Douglas et al., 1986). development of criminal profiles today. The
Disorganized scenes, in contrast, exhibited a present study uses a broader categorization,
form of murder carried out by an offender who such as those designed by Holmes and Holmes
was less apt to plan the offence, who obtained (1998). The present study includes influencing
victims by chance, and who behaved haphaz- factors before the kill, such as personal histor-
ardly during the crime (Douglas et al., 1986). ies and serial killers’ experiences of abuse.
This original typology, that of the organ-
ized or the disorganized offender, was deemed
overly simplistic and has since broadly Abuse
expanded (Canter, 1994; Holmes & Rossmo, The World Health Organization (WHO)
1996; Turco, 1990). Recently, researchers defines child abuse as “all forms of physical
have developed more sophisticated typologies and/or emotional ill-treatment, sexual abuse,
including (1) visionary, mission-oriented, neglect or negligent treatment or commercial
hedonistic, and power-control oriented killers or other exploitation, resulting in actual or
(Holmes, De Burger, & Holmes, 1988); (2) potential harm to the child” (World Health
thrill-motivated killers, murders for profit, and Organization, 1999, p. 80). Although this def-
family slayings (Levin & Fox, 1985); and (4) inition covers a spectrum of abuse, the three
travelling serial killers, local serial killers, and main types defined in the present study are
Comparing Younger and Older Adult Eyewitnesses 3
episodes between people (Beale, Cox, Clarke, (n ¼ 35), physical abuse (n ¼ 36), sexual
Lawrence, & Leather, 1998; Taylor et al., abuse (n ¼ 21), psychological and physical
2017), and marital conflict (Gottman, 1979). abuse (n ¼ 88), physical and sexual abuse
(n ¼ 7), and physical, sexual, and psycho-
logical abuse (n ¼ 46). The dates of first kill
Present study ranged from 1850 to 2014. The date of last kill
The present study uses a BSA approach to ranged from 1893 to 2014. In calendar years
investigate the effects of different types of of the sample at the time of their first kill
early childhood abuse (physical, psycho- ranged from 6 to 60 (M ¼ 28, SD ¼ 8.96), and
logical, and sexual abuse) on later serial kill- their last kill ranged from 16 to 68 (M ¼ 34,
ings. The pattern of actions explored begins SD ¼ 10), although the exact age in childhood
with this early abuse, leading on to the typ- when abuse occurred is unknown. The number
ology of the serial killer. This is included in of kills ranged from 3 to 138 within several
the analysis to indicate links between abuse countries: Brazil (n ¼ 3), Canada (n ¼ 6),
and typology, rather than direct sequential Australia (n ¼ 6), USA (n ¼ 176), Argentina
effects.1 The effect of experiencing multiple (n ¼ 1), Columbia (n ¼ 4), Ecuador (n ¼ 1),
types of abuse at the same time was also inves- England (n ¼ 8), France (n ¼ 4), Germany
tigated. Typologies were classified into four (n ¼ 4), Italy (n ¼ 1), Mexico (n ¼ 2) Ireland
groups, dependent on the serial killer’s motiv- (n ¼ 1), Scotland (n ¼ 1), Pakistan (n ¼ 1),
ation: lust, anger, power, and financial gain. Russia (n ¼ 5), South Africa (n ¼ 7), and
The next behaviour explored was the crime Spain (n ¼ 2). As the sample had been
scene behaviour – such as how the victim was obtained from secondary sources and so does
killed and what was done with the body. Thus, not contain any studies with human partici-
the sequence from early childhood abuse, typ- pants, ethics approval was not needed.
ology of the killer, and crime scene behaviours
was analysed. While formal hypotheses are Coding procedure
not made, owing to the novel nature of the The sample was split according to the type of
research, several expected links can be out- abuse experienced in childhood. The typology
lined. First, it is likely that childhood sexual of the serial killer in each group was then
abuse will lead predominantly to sexual typol- coded (Lust/rape, power, financial gain, or
ogies, taking into account previous literature anger) into the BSA. Lust/rape killers were
highlighting that violent upbringings influence those whose murders involved sexual ele-
later delinquency, adult criminality, and vio- ments, including rape, sexual assault without
lence (Maxfield & Widom, 1996). It is also penetration, or symbolic sexual assault such as
likely that individuals who have experienced the insertion of a foreign object into body orifi-
early physical abuse will show a greater ces (Douglas et al., 1992). Power killers were
amount of violence – for example, signs of tor- those who derived pleasure from having com-
ture and overkill (infliction of excessive and plete control over their victims. Financial gain
unnecessary violence). killers were those who killed for motivations
based on the accumulation of goods or finan-
ces. Anger killers were those who killed for
Methods
motivations that stemmed from feelings of
Sample anger, frustration, or betrayal, whether real or
An all-male sample of 233 serial killers with a imagined. The overall methods used across
documented history of childhood abuse was kills was also recorded for all killers.2 The
collected. Numbers experiencing each type of final factor considered for each individual was
abuse were as follows: psychological abuse what they had done with their victim’s body(s)
Comparing Younger and Older Adult Eyewitnesses 5
after the murder (e.g., moved the body to a dif- typologies. Rape/lust typology was the most
ferent location and buried it, hid the body at common typology in the current dataset
the crime scene, etc.). Percentages of partici- (n ¼ 152). However, it had followed more fre-
pants for each individual were calculated at quently from psychological abuse (n ¼ 10,
each stage. SR ¼ 7.06) and a combination of all three
A coding scheme was developed based on types of abuse (n ¼ 12; SR ¼ 7.04). There did
every recorded outcome/behaviour reported in not seem to be a strong connection between
the dataset. Given the straightforward nature financial gain and any type of abuse or com-
of the task, there was no ambiguity over bination of abuse, as it was infrequent in all
responses or coding. The typology of serial cases, particularly the experience of all three
killers was assessed by forensic psychologists. combined. There was no strong pattern
between any single type of abuse and anger
Statistical analysis typology, and only 23 subjects classified as
After data were coded into chains of discrete this typology. There was a clear pattern
behaviours and categories, data were entered between rape/lust typology and torture of the
into the statistical software R (R Core Team, victim (n ¼ 16, SR ¼ 9.59). There was a clear
2013) and analysed using a behaviour pattern between financial gain and the murder
sequence analysis program developed by the being carried out quickly (n ¼ 12, SR ¼ 8.33).
researchers. The program calculated frequen- An additional benefit of the BSA approach is
cies of individual behaviours, transitional fre- that particular cases can be highlighted and
quencies, chi-square (v2) statistics, and analysed individually. For instance, the follow-
standardised residuals. ing analyses focused on each type of abuse
sequence by itself. This allows researchers and
investigators to refine their search parameters
Results and to begin narrowing in on particular
The main stage of BSA is to focus on the tran- sequences based on evidence or interests.
sitions between pairs of behaviours. Transition
frequencies between antecedents and sequiturs Physical abuse
are calculated, and chi-square analyses are For physical abuse (Figure 2), there was a dis-
indicated if these transitions occur above the
tinct sequence between the experience of phys-
level of chance. State transition diagrams can
ical abuse and rape/lust typology (n ¼ 6,
then be drawn, which indicate pairs of behav-
SR ¼ 4.80) and anger typology (n ¼ 2,
iours with high standardised residuals (SR). It
SR ¼ 2.77). Those with rape/lust typologies
is important to note that while pairs of behav-
were more likely to carry out the murder
iours can be connected to form longer chains,
quickly (n ¼ 5, SR ¼ 5.75), and crime scenes
the analyses are only on pairs of behaviours.
exhibited signs of the victim having been
Longer chains, though intuitively appealing,
bound (n ¼ 3, SR ¼ 4.64). There was also evi-
are actually limited in terms of generalizabil-
dence of overkill, and in all cases where over-
ity, owing to over-fitting of data. All of the
kill occurred, the body had been left at the
transition lines in the diagram are significant
crime scene.
(p < .05) (see Figure 1).
Importantly, Figure 1 shows that there is a
clear distinction between type of abuse experi- Psychological abuse
enced and later typology of the serial killer. For psychological abuse (Figure 3), there was
For example, experiences of sexual abuse a distinct sequence between the experience of
most likely to lead to the power typology psychological abuse and rape/lust typology
(n ¼ 6, SR ¼ 9.21), compared to other (n ¼ 10, SR ¼ 6.50) and financial gain (n ¼ 5,
6 A. Marono et al.
Figure 1. State transition diagram for type of abuse experienced, typology, and crime scene behaviour.
Standardised residuals indicated by line thickness (see key).
SR ¼ 4.60). Murders were carried out quickly typology was more strongly related to carry-
in all cases where the motivation was financial ing out the murder quickly (n ¼ 4, SR ¼ 5.23)
gain; however, if the typology was rape/lust, than to mutilating the victim (n ¼ 2,
then fewer were carried out quickly (n ¼ 5, SR ¼ 3.16). The anger typology showed a link
SR ¼ 3.63). There was also a strong link to carrying out the murder quickly (n ¼ 2,
between torture and evidence of overkill SR ¼ 2.16), and financial gain was also linked
(n ¼ 5, SR ¼ 6.25), and evidence of overkill to carrying out the murder quickly (n ¼ 4,
and mutilation of the body (n ¼ 6, SR ¼ 6.84). SR ¼ 5.85). Finally, there were strong links
between torturing and binding the victim
Sexual abuse (n ¼ 4, SR ¼ 9.39), and the victim being
bound and mutilation (n ¼ 4, SR ¼ 8.63).
Unlike the other types of abuse, sexual abuse
was linked to all four typologies (see
Figure 4). The rape/lust typology was slightly All types of abuse
more likely to the torture of victims (n ¼ 4, When a combination of sexual, physical and
SR ¼ 5.92) compared to showing signs of psychological abuse was experienced
overkill (n ¼ 4, SR ¼ 4.27). The power (Figure 5), the rape/lust typology for killing
Comparing Younger and Older Adult Eyewitnesses 7
Figure 2. State transition diagram for physical abuse, typology, and crime scene behaviour.
Standardised residuals indicated by line thickness (see key).
Figure 3. State transition diagram for psychological abuse, typology, and crime scene behaviour.
Standardised residuals indicated by line thickness (see key).
was most likely to follow (n ¼ 12, SR ¼ 6.76). anger were more likely to carry out the murder
Rape/lust was more likely to lead to the vic- quickly (n ¼ 4, SR ¼ 4.49) than show evidence
tims being bound (n ¼ 6, SR ¼ 7.11) than to of overkill (n ¼ 2, SR ¼ 1.61).Those who
killers carrying out their murders quickly killed for financial gain carried out the murder
(n ¼ 4, SR ¼ 2.63). Subjects who killed for quickly (n ¼ 3, SR ¼ 5.03).
8 A. Marono et al.
Figure 4. State transition diagram for sexual abuse, typology, and crime scene behaviour. Standardised
residuals indicated by line thickness (see key).
Figure 5. State transition diagram for typology and crime scene behaviour following experience of psy-
chological, physical, and sexual abuse. Standardised residuals indicated by line thickness (see key).
victims, and how murders were carried out. therefore, no unnecessary means of killing,
The outcome is an insight into the sequential infliction of unnecessary pain, or evidence
chains that different types of abuse result in of enjoyment.
for an individual. Within the current dataset, Those who were classified as exhibiting
results indicate that different types of abuse rape/lust typology commonly engaged in post-
affect later typologies and murder behaviours. mortem sex, regardless of the type of abuse
Previous literature suggests that early experienced as a child. There was also no evi-
physical abuse leads to later aggression and dence of overkill in any of the cases, although
violence (Widom, 1989). Current results par- torture was commonly used. A possible
tially supported this. Although those who were explanation for this is the presence of abnor-
physically abused were more likely to demon- mal paraphilias or sexual sadism, which sup-
strate ‘overkill’ of their victim, the most spe- ports Dietz, Hazelwood, and Warren’s (1990)
cifically violent methods of kill were practised argument that psychopathic sexual sadists kill
by those who had been sexually or psycho- for the sheer pleasure of torturing and murder-
logically abused in early life. For example, ing their victims in a sexual way. Importantly,
mutilation, torture, and binding the victim the experience of sexual abuse, whether iso-
were more typical of serial killers who had lated or experienced alongside physical and/or
experienced sexual abuse. Furthermore, those psychological abuse, led to the mutilation or
who had been sexually abused rarely showed torture of the victim. Similarly, individuals
evidence of overkill, and the murders tended who classified as lust/rape typology where
to be carried out quickly. This was not the more likely to torture or mutilate their victims.
case for both physical and psychological This suggests a correlation with sexual behav-
abuse, as both showed evidence of overkill. iour and a need to inflict pain.
The exact reason for this cannot be clarified Additionally, results are incongruent with
from this sequence chain, although based on previous literature on typologies, as there was
previous research (Briere & Elliott, 1994; no consistent pattern for method of killing and
Wyatt & Newcomb, 1990), it may be that disposal of the body within each typology.
these patterns emerge because those who have This supports the suggestion by Canter and
experienced sexual abuse suffer from deep- Wentink (2004) that features of power/control
seated anger and self-blame, leading them to typologies were consistent for serial killers
lash out and kill their victims quickly, and rather than forming a distinct type. Thus, the
they are more likely to feel guilt or remorse reliability of isolated typologies is less mutu-
afterwards and thus are unlikely to show evi- ally exclusive than previously believed, and
dence of overkill. more attention should be paid to what factors
Furthermore, all recorded murders were influence specific methods of killing than to
carried out quickly by those who were classi- the motivations of individual offenders.
fied as motivated by power. There was also Indeed, it may be that the cross-sectional
no evidence of any torture, mutilation, or over- approach to typology defining could be devel-
kill. Although again, the sequence chain can- oped to include temporal dimensions. The cur-
not direct determine the reason as to why this rent analytical method can be used to show
is, based on previous research (Canter & linkages between behaviours and events, over
Wentink, 2004; Holmes & Holmes, 1998), a time, which may provide investigators with a
reason could be that this stems from a need to clearer understand and method for develop-
control the victim and assert dominance. ing typologies.
Therefore, killers see the act of killing as a A limitation of the current research is the
necessity, rather than obtaining any enjoyment potential influence of additional life events
out of the kill, per se. In these killings, there is, that may intervene in the current diagrams, as
10 A. Marono et al.
Fossi, J.J., Clarke, D.D., & Lawrence, C. (2005). Keatley, D.A. (2018). Pathways in crime: An
Bedroom rape sequences of sexual behavior introduction to behaviour sequence analysis.
in stranger assaults. Journal of Interpersonal London: Palgrave MacMillan.
Violence, 20(11), 1444–1466. doi:10.1177/ Lawrence, C., Fossi, J., & Clarke, D. (2010). A
0886260505278716 sequential examination of offenders’ verbal
Gottman, J.M. (1979). Marital interaction: strategies during stranger rapes: the influence
Experimental investigations. New York, NY: of location. Psychology Crime and Law,
Academic Press. 16(5), 381–400. doi:10.1080/1068316090
Grubin, D., Kelly, P., & Ayis, S. (1997). Linking 2754964
serious sexual assaults. Mahipalpur, India: Levin, J., & Fox, J.A. (1985). Mass murder:
Home Office, Police Policy Directorate, America’s growing menace (pp. 3–7). New
Police Research Group. York, NY: Plenum Press.
Hazelwood, R.R., & Douglas, J.E. (1980). The Marono, A., Clarke, D., Navarro, J., & Keatley,
lust murderer. FBI Law Enforcement D. (2018). A sequence analysis of nonverbal
Bulletin, 49, 18. behavior and deception. Journal of Police
Hazelwood, R.R., & Warren, J.I. (2000). The and Criminal Psychology, 33(2), 109–117.
sexually violent offender. Aggression and doi:10.1007/s11896-017-9238-9
Violent Behavior, 5(3), 267–279. doi:10. Marono, A., Clarke, D.D., Navarro, J., &
1016/S1359-1789(99)00002-6 Keatley, D.A. (2017). A behaviour sequence
Hewitt, A., Beauregard, E., & Martineau, M. analysis of nonverbal communication and
(2016). Can body disposal pathways help the deceit in different personality clusters.
investigation of sexual homicide? In E. Psychiatry, Psychology and Law, 25(5),
Beauregard & M. Martineau, (Eds.), The 730–744. doi:10.1080/13218719.2017.1308
sexual murderer: Offender behaviour and 783
implications for practice. Abingdon, UK: Maxfield, M.G., & Widom, C.S. (1996). The
cycle of violence: Revisited 6 years later.
Routledge.
Archives of Pediatrics and Adolescent
Holmes, R.M., De Burger, J., & Holmes, S.T.
Medicine, 150(4), 390–395. doi:10.1001/
(1988). Inside the mind of the serial murder.
archpedi.1996.02170290056009
American Journal of Criminal Justice, 13(1),
Miller, L. (2014). Serial Killers I: Subtypes, pat-
1–9. doi:10.1007/BF02890847
terns, and motives. Aggression and Violent
Holmes, R.M., & Holmes, S.T. (Eds.). (1998).
Behavior, 19(1), 1–11. doi:10.1016/j.avb.
Selected problems in serial murder investiga- 2013.11.002
tions. In Contemporary perspectives on serial Mitchell, H., & Aamodt, M.G. (2005). The
murder, 227–234. Thousand Oaks, CA: incidence of child abuse in serial killers.
Sage. doi:10.4135/9781452220642.n16 Journal of Police and Criminal
Holmes, R.M., & Rossmo, D.K. (1996). Psychology, 20(1), 40–47. doi:10.1007/
Geography, profiling, and predatory crimi- BF02806705
nals. Profiling violent crimes. Thousand R Core Team. (2013). R: A language and envir-
Oaks, CA: Sage. onment for statistical computing. Vienne,
Ioana, I.M. (2013). No one is born a serial killer. Austria: R Foundation for Statistical
Procedia - Social and Behavioral Sciences, Computing, ISBN 3-900051-07-0, Retrieved
81, 324–328. doi:10.1016/j.sbspro.2013.06. from http://www.R-project.org/
436 Reid, S. (2016). Compulsive criminal homicide:
Joiner, T. (2007). Why people die by suicide. A new nosology for serial murder.
Cambridge, MA: Harvard University Press. Aggression and Violent Behavior, doi:10.
Keatley, D.A., Barsky, A.D., & Clarke, D.D. 1016/j.avb.2016.11.005
(2016). Driving under the influence of alco- Ressler, R.K., Burgess, A.W., & Douglas, J.E.
hol: a sequence analysis approach. (1988). Sexual homicide: Patterns and
Psychology, Crime and Law, 23(2), motives. New York, NY: Simon and
135–146. doi:10.1080/1068316X.2016. Schuster.
1228933 Ressler, R.K., Burgess, A.W., Douglas, J.E., &
Keatley, D.A., Golightly, H., Shephard, R., Depue, R.L. (1985). Violent crime. FBI Law
Yaksic, E., & Reid, S. (2018). Using behav- Enforcement Bulletin, 54(8), 2–31.
ior sequence analysis to map serial killers’ Taylor, O., Keatley, D.A., & Clarke, D.D.
life histories. Journal of Interpersonal (2017). A behavior sequence analysis of per-
Violence. doi:10.1177/0886260518759655 ceptions of alcohol-related violence
12 A. Marono et al.