Using Behavior Sequence Analysis to Map Serial Killers’ Life Histories

Abstract

The aim of the current research was to provide a novel method for mapping the developmental sequences of serial killers’ life histories. An in-depth biographical account of serial killers’ lives, from birth through to conviction, was gained and analyzed using Behavior Sequence Analysis. The analyses highlight similarities in behavioral events across the serial killers’ lives, indicating not only which risk factors occur, but the temporal order of these factors. Results focused on early childhood environment, indicating the role of parental abuse; behaviors and events surrounding criminal histories of serial killers, showing that many had previous convictions and were known to police for other crimes; behaviors surrounding their murders, highlighting differences in victim choice and modus operandi; and, finally, trial pleas and convictions. The present research, therefore, provides a novel approach to synthesizing large volumes of data on criminals and presenting results in accessible, understandable outcomes.

Keywords

serial killer homicide violent crime behavior sequence analysis profiling

Serial killing is a violent, extreme, and relatively rare form of human crime behavior (Allely, Minnis, Thompson, Wilson, & Gillberg, 2014; Hodgkinson, Prins, & Stuart-Bennett, 2017). In fact, serial homicide is so rare that it is estimated to account for less than 1% of all homicides. However, despite its infrequency, it is important to investigate and understand the antecedents of this serious crime for those affected by it. Serial homicide is a difficult term to define, with several variations being suggested; however, the Federal Bureau of Investigation (FBI) currently defines it as “the unlawful killing of two or more victims in separate occasions” (FBI, 2008, p. 12). Serial homicide is differentiated from other forms of multiple-victim homicides (e.g., mass murder or spree killing), on the basis of several features, including the number of murder events and the fact that serial killers, unlike mass and spree killers, carefully select their victims and the location of the murders (Douglas, Burgess, Burgess, & Ressler, 1992). And, recently, Reid (2017) clarified the term “serial homicide” by redirecting attention from the number of victims killed to the compulsive, psychological aspects of the criminals themselves.

Researchers have faced significant challenges in narrowing down the definition of serial homicide. However, a greater difficulty has been predicting who will commit such crimes, especially given the infrequency of the crime and the limitation of not all facts being known or available. Recent advancements in serial homicide datasets and the development of new methodological approaches for studies in risk assessment, violence prevention, and identification have furthered research efforts toward predictive modeling (Ioana, 2013; Miller, 2014). For example, Simkin and Roychowdhury (2014) used stochastic analyses to determine the minimum number of days between each murder committed by serial killer Andrei Chikatilo. And, recently, Bayesian models have been used as a way to approach criminal geographical profiling among serial offenders (Leitner & Kent, 2009). While models for detection and apprehension are beginning to advance, and datasets continue to grow, however, one area of research interest seems to have gone relatively unexplored. In particular, our knowledge of the ways in which developmental risk factors occur and impact developmental trajectories that lead to serial homicidal offending has yet to be fully explored.

Risk Factors

Researchers have shown a number of factors that are present in many serial killers’ life histories. For example, in his review of nearly 100 full-length biographies of serial murderers, Stone (2001) found parental brutality (i.e., excessive mistreatment of a child) to be present in 54% (n = 49) of cases. In another study, Ressler, Burgess, and Douglas (1988) found evidence of frequent moving during childhood in one third of serial killers. Ressler et al. (1988) also observed that more than 40% of serial killers lived outside of their familial homes by the age of 18. In terms of biological mediators, head injury is believed to be present in upward of 10% of serial killers, as, too, is Autism Spectrum Disorder (ASD; Allely et al., 2014). Investigations into the parent-child relationship between mothers and serial killers have revealed relationships typified by abuse. For example, studies have frequently shown mothers of serial killers to have been pervasively rejecting, punitive, hateful, smothering, controlling, and/or infantilizing (Hickey, 1992; Miller, 2014). Also, Canter, Missen, and Hodge (1996) analyzed the backgrounds of serial killers and found that 75% had been previously convicted of crimes. These findings have been replicated by Harbort and Mokros (2001) who found that 79% of serial killers had previous convictions, whereas in the study by Canter et al. (1996), nearly half had been arrested as juveniles.

Large volumes of research have outlined the role of childhood experiences as they relate to the serial killer. This has provided researchers with an important foundation for understanding the early childhoods of serial killers and has subsequently provided researchers and investigators with some foundation of the types of behaviors and events to focus on when conducting their own analyses. However, the issue remains that despite having this knowledge, we have yet to address how biological, social, and psychological features operate and interact to shape the developmental trajectories of serial killers.

Our current knowledge regarding the developmental risk factors for serial homicide is primarily informed by outdated statistics drawn from small, nonrepresentative samples, and analyzed by way of simple descriptive methods including frequency and distribution. Beyond the useful information that simple quantitative designs have offered, these approaches contribute more of a general checklist of features found in the backgrounds of serial killers rather than an in-depth investigation into the directional relationships or temporal significance of these findings. Thus, while we have built a strong foundational literature regarding the developmental risk factors that are associated with serial homicide, the utility of this information ends at questions that address a simple “yes or no” or “is x correlated to y.”

Because the inferences we can draw from relatively simple methodological designs are fairly limited, there is currently a significant knowledge gap in the developmental literature on serial homicide. For example, we know very little about how the temporal order of these risk factors may influence later serial killers’ behaviors, and whether the temporal order gives a clearer map of why individuals develop into serial killers. Given the need for increased complexity in the literature on the serial killer developmental serial, we seek to understand the effect of temporal order on development using an advanced-method statistical model, behavior sequence analysis (BSA).

BSA

BSA, also referred to as Lag-Sequential Analysis (LSA) or simply Sequence Analysis, is a useful method for mapping and statistically highlighting significant transitions between behaviors and events (Bakeman & Gottman, 1986; Clarke & Crossland, 1985; Keatley, Barsky, & Clarke, 2017; O. Taylor, Keatley, & Clarke, 2020; P. J. Taylor et al., 2008). Rather than studying the frequency of individual risk factors, as outlined above, and perhaps performing statistical tests on these factors as individual predictors or groups, BSA, instead, focuses on the transitions between these factors. For instance, rather than analyzing behaviors A, B, and C as a group of risk factors, BSA, instead, analyzes whether A is an antecedent of B, and B is an antecedent of C (effectively: A → B, B → C, etc.). This approach, therefore, has temporal dynamics at its core, and can be used to understand developmental trajectories. In addition, interactions between serial killers and other people (e.g., families, victims, legal systems) can also be mapped. Indeed, a strength of the BSA approach is providing very clear life maps, called state transition diagrams, which outline significant transitions between events and behaviors (Ellis, Clarke, & Keatley, 2017; Marono, Clarke, Navarro, & Keatley, 2017, 2018).

In BSA, chains of events (i.e., risk factors in serial killers’ lives) are chronologically ordered, based on mathematical Markov models (Ivanouw, 2007). This approach can either be theoretically driven (i.e., looking for and coding known risk factors), or it can take a more data-driven, bottom-up approach (i.e., recording all events in a serial killers’ life, rather than only those that fit predefined checklists). Identifying only those behaviors that fit with predetermined checklists runs the risks of missing valuable events and behaviors that may be new or not otherwise previously noted. Therefore, while taking longer, using a more inclusive, bottom-up approach means that no event or behavior is dismissed at the early coding stage. Theoretical knowledge can then be applied to data to see which models fit with these data and are supported.

Sequence Analysis research has previously been used in a number of forensic areas, such as rape (Ellis et al., 2017; Fossi, Clarke, & Lawrence, 2005; Lawrence, Fossi, & Clarke, 2010), road traffic accidents and crimes (Keatley et al., 2017), deception and body language (Marono et al., 2017, 2018), and violence in nighttime venues (O. Taylor et al., 2020). This research typically uses a lag-one sequence analysis approach. This means that behaviors are analyzed in pairs (i.e., A ≥ B, B ≥ C, C ≥ D). Longer sequence methods (i.e., AB ≥ C) are possible, but for a number of reasons are usually avoided in applied research (see Ellis et al., 2017).

Present Research

The aim of the current research was to show the benefits of using a BSA method for understanding serial killers’ life histories. This approach allows for very in-depth, qualitative accounts of life histories to be systematically quantified and analyzed with traditional statistical tests (see Keatley et al., 2017). Typically, BSA research focuses in on rather brief time periods; however, BSA can also be used to look at large-scale time frames. The trade-off can be likened to reading a map, wherein readers can “zoom in” on very specific, small-scale detail, or “zoom out” and see more global structures. Therefore, less information is given when looking at life histories, compared with a 1-hr episode; however, “larger” life structures can still be observed.

Although the current research used a bottom-up approach to data collection and coding, it is not completely atheoretical,¹ and based on previous research, a number of hypotheses can be formed. First, well-known risk factors (e.g., social and economic deprivation, life stressors, history of legal encounters) are expected to be present in the final sequence analysis outcomes. What is not known, however, is how these risk factors will be temporally ordered. Therefore, the current research can be used to complement and support existing findings; however, it will add the temporal sequence to known risk factors.

Method

Sample

Biographical and autobiographical data were collected for 25 serial killers from a combination of online sources, books, medical reports (when available), psychiatric reports, and police reports. Inclusion criteria were based on access to multiple sources of verifiable data, so that cross-checks and validity of facts could be established. An original list of 60 serial killers was first collected, and from this, the 25 with the most information available to researchers were used for the final analysis. All information included in the final research had to be verified by at least three sources for it to be included.²

The sample of serial killers included in the current research varied across gender (female = 7, male = 18), nationality (White American = 10, White British = 6, African American = 4, White European = 2, Hispanic = 2, and Asian = 1), and number of victims (see Table 1). Of the 25 serial killers included, the average age at first kill for this sample was 31.12 (SD = 8.99, range = 18-53). The average number of victims for this sample ranged from three to 215. The majority (36%) of this sample had less than 10 victims, 12% had more than 100 victims. Obviously, it is very hard to establish an actual number of victims, so estimates were calculated based on cross-validation of sources.

Table 1.

Descriptive Information for the Sample of Serial Killers.

Name	Also Known As	Race	Gender	Place of Birth	Age at First Kill	No. of Victims^a	Time Period
Harold Shipman	The Killer Doctor	WB	M	England	24	>100	1970-1998
Aileen Wuornos	NA	WA	F	Michigan	33	<10	1989-1990
Anna Marie Hahn	Arsenic Anna, Female Bluebeard, The Blonde Borgia	WE	F	Germany	26	<10	1932-1937
John Wayne Gacy	The Killer Clown	WA	M	Chicago	30	>30	1972-1978
Peter Sutcliffe	The Yorkshire Ripper	WB	M	Yorkshire	29	>10	1975-1980
Steve Wright	Suffolk Strangler, Ipswich Ripper	WB	M	Norfolk	48	<10	2006
John Christie	The Whispering Killer, The Rillington Place Strangler	WB	M	Yorkshire	45	<10	1943-1953
Dorothea Puente	Death House Landlady	WA	F	California	53	<10	1982-1988
Dennis Rader	BTK Killer	WA	M	Kansas	29	>10	1974-1991
Dennis Nilsen	Muswell Hill Murderer, The Kindly Killer	WB	M	Scotland	33	>10	1978-1983
Jeffrey Dahmer	Milwaukee Cannibal	WA	M	Milwaukee	18	>10	1978-1991
Anthony Sowell	The Cleveland Strangler	AA	M	Cleveland	48	>10	2007-2009
Henry Lee Lucas	The Confession Killer	WA	M	Virginia	24	<10	1960-1983
Lonnie David Franklin Jr.	The Grim Sleeper, Southside Slayer	AA	M	California	33	>10	1985-2007
Pedro Alonso Lopez	Monster of the Andes	H	M	Colombia	21	>100	1969-1980
Lorenzo Gilyard Jr.	The Kansas City Strangler	AA	M	Missouri	35	>10	1977-1993
Luis Garavito	The Beast	H	M	Colombia	35	>100	1992-1999
Henry Louis Wallace	The Taco Bell Strangler	AA	M	S. Carolina	25	<10	1990-1994
Beverley Allitt	Angel of Death	WB	F	England	23	<10	1991
Jane Toppan	The Nightmare Nurse	WA	F	Massachusetts	30	>30	1887-1901
Belle Gunness	Lady Bluebeard, Mistress of Murder Farm, Hell’s Belle	WE	F	Norway	25	>30	1884-1908
Harvey Glatman	The Lonely Hearts Killer, The Glamour Girl Slayer	WA	M	New York	30	<10	1957-1958
Nannie Doss	The Jolly Black Widow, The Giggling Grandma	WA	F	Alabama	22	>10	1927-1954
Ted Bundy	NA	WA	M	Vermont	27	>30	1973-1978
Yang Xinhai	The Monster Killer	A	M	China	32	>50	2000-2003

Note. Race is coded as WB (White British), WA (White American), WE (White European), AA (African American), H (Hispanic), A (Asian). Gender is coded as M (male) and F (female). Number of victims is coded as <10 (less than 10), >10 (more than 10), >30 (more than 30), >50 (more than 50), and >100 (more than 100).

Number of victims are approximates based on information from sources.

Although the current research did not involve participants, for the sake of completion, ethical approval was still obtained for the current study by the Ethics Committee at the University of Lincoln.

Coding

Before analyzing serial killers’ life history accounts, a list of behaviors and events were written, based on journal searches of “risk factors” and “developmental factors” in serial killers’ lives. Several articles outlined the associated risk factors in early life for serial killing. Other research outlined behaviors that commonly occur during the crime, such as a sexual element, fetishes or paraphilia, criminal history, and intrinsic motivation (Arndt, Hietpas, & Kim, 2004; Harbort & Mokros, 2001; Mott, 1999). As outlined, to ensure as many events and behaviors were included as possible, additional behaviors were added by the researchers during the course of data collection and coding as they were identified in the (auto-)biographical data. A final total of 199 behaviors were identified across the current sample.

Once the final list of codes had been developed, they were used to recode biographies into sequential chains (similar to the method used in Keatley et al., 2017). Essentially, long paragraphs and pages of autobiographical information is reduced to abbreviated codes for particular behaviors and events. These code sequences were then checked to ensure that no essential information was either missing or made ambiguous. Once two researchers had agreed on the coding of a serial killer’s life, the code was entered into the database. No disputes arose in the current dataset, therefore, no further discussion was needed on coding.

Statistics

Once serial killers’ lives were coded into sequences of behaviors and events, transition frequencies were calculated (essentially, how many A →; B, B →; C, C →; D pairs occurred). These frequencies were put into a transition frequency matrix, which is a contingency table showing which sequiturs follow each antecedent. This was calculated across all serial killers in the sample, and once the table was calculated, chi-square statistic was used, which provided Standardized Residuals (SR). SRs indicate which cells in the transition frequency matrix are occurring above the level of chance—essentially, which transitions occur more than would be expected by chance in these data.

Results

The first stage of BSA is to identify the most frequently occurring events. This relates closely to previously established research and goes some way to show how the current method complements existing knowledge. Once the frequencies of behaviors and events have been analyzed, the main part of BSA is to calculate the significant transitions between behaviors and events. Significant transitions are then shown in the state transition diagram, which presents the temporal order of events, from childhood to later life events, including murder, arrest by police, and court trials and outcomes.

Frequencies

The first part of the analysis readily relates to previously known and established findings on key risk factors in serial killers’ lives. In a sample of 25 serial killers, high frequencies of “born,” “arrest,” and “trial” are to be expected, as all life histories begin at birth, and a trial commonly follows an arrest. As identified in Table 2, other high-frequency events include “married,” “divorced,” and “had a child/children,” which are also common events to occur within an individual’s lifetime. A list of the most frequently coded behaviors is presented in Table 2. Some of these findings, though expected (i.e., “arrested”), occur a higher proportion of times than one might initially expect. For example, although in the current sample “arrested” is an expected and likely behavior for any crime research, including serial killing, “arrested” occurs 49 times in a sample of 25, and indicates a high rate of arrests among serial killers.

Table 2.

Frequencies of Behaviors and Events.

Behavior/Event	Frequency
Arrested	49
Murders went undetected	37
Continue killing	28
Consistent M.O.	26
Moved location; Born	25
Trial; First kill	24
Found guilty	23
Murdered <10 victims	22
Committed crimes of sexual assault; Access to victims	19
Married	16
Divorced	15
Went to prison; Served sentence; Ligature strangulation; Had a child/children; Changed M.O.; Bodies buried/disposed of	14
Rape	13
Remarried; Crimes of theft	12
Received death sentence; Poisoned (overdose); New job; Gained confession	11
Received life sentence; Dropped out of school; Discharged; Attended medical school	10
Paroled or given early release; No sentence/charges/action; Attempted murder; Abusive childhood	9
Take photos; Searched house; Lure victims to home; Found evidence in house; Final victim led to arrest; Enlisted in the armed forces; Alienated from peers; Alcoholic family member(s)	8
Stabbing/slitting; Restrain victims; Impoverished/lived in poverty; Executed; Enter victim’s home; Break-in killing; Alcoholic/heavy drinking	7
Widowed; Sexually assaulted by other (nonfamilial); Serving prison sentence; Job provided access to victims; Institutionalized; Fired from job/left job; DNA evidence; Dismember/destroy corpse; Disciplinarian/strict/religious household; Could not keep a job; Benefited financially; Beaten to death	6
Victim body found; Use sex worker; Targeted female victims; Rejected; Questioned by police; Plead not guilty; Partner cheated; Murder interrupted; Kept souvenirs for victims; Injured or sickly as a child; Incapacitate victims; Father died; Exhumation/autopsy of victim; Domestic violence; Died in prison; Developed drug addiction; Described murders (with detail); Bodies kept in the house; Attempted suicide; Arrested for a different crime	5
Strangulation; Shot/shooting; Sexual acts involving corpse; Selected isolated location; Normal childhood; Murdered >10 victims; Interest in dead bodies; Hired for another job; Force them to chosen location; Cover up crimes; Consumption of drink/drugs; Committed crimes of assault; Committed crimes of burglary; Began working in medical field	4
Victims became unspecific (whoever was encountered); Unpopular with peers/people; Targeted prostitutes; Targeted male victims; Targeted African American women; Taken in/adopted by others; Sought recognition for kills; Sexual deviance; Revisited crime scene; Received medical attention; Popular with peers/people; Plea bargain/reduced sentence; Parents divorced/separated; On death row; Murdered >30 victims; Murdered >100 victims; Mother died; Lure victims to isolated location; Left home; Kicked out of home; Immigrated (to United States); Found religion; Father left; Elements of theft in murder; Committed miscellaneous crimes; Committed crimes of fraud; Bodies left in secluded locations/concealed; Bodies dumped; Acts involving animals (torture/sexual)	3
Working-class family; Witness testimony; Went to university; Went to police; Vanished/disappeared/unknown; Targeted male children (under 20); Targeted family members; Sexually assaulted by other (family); Sex with victim; Scapegoat; Provide care for victim; Pose victim corpses; Police investigation; Plead guilty; Mother left; Interest in the effect of drugs; Insanity plea; Incestuous relationship; Father in prison; Family members in prison; Escaped; Elements of fraud in murder; Elements of arson; Drifted; Detained under Mental Health Act; Deported; Defense of mercy killings; Conflicted over sexuality; Committed crimes of arson; Committed crimes involving weapons; Bodies exposed; Attempts on life while in prison; Alter medical notes; Cleaned corpse	2
Voyeuristic; Unhealthy interest in prostitutes; Unable to conceive; Teenage pregnancy; Targeted the elderly; Targeted patients; Targeted infant patients; Targeted elderly patients; Targeted elderly male patients; Targeted African American men; Target known females; Target female children (under 20); Sex worker; Sentence commuted to life in prison; Retracted claims; Pregnant (unmarried); Pled diminished responsibility; Picked up by victims; Overprotective mother; Other died (family); Murdered >50 victims; Mother ill; More victims uncovered; More suspected victims; Killed in prison; Harmful behaviors; Fake illness; Experimenting with corpse disposal; Elements of sexual assault in murder; Elements of sexual arousal; Elements of cannibalism; Disposal of bodies changed because of limits in space; Committed crimes of armed robbery; Claimed self-defense; Child adopted; Attempted to flee; Addictions (other)	1

Note. Behaviors/events with the same frequency have been put into the same rows. M.O. = modus operandi.

BSA

BSA was conducted to show the progression of events across the sample of serial killers’ lives, from birth, through murders, to trials and outcomes. A transition frequency matrix was calculated (see Supplementary Material), which shows the frequency of transitions between events. SRs were also calculated for each transition, which indicated the extent to which the transition occurred, compared with chance; effectively, whether “B” followed “A” more times than would be expected by chance.

A total of 171 significant (p < .05) transitions were identified in the current dataset, which results in an extremely complex diagram (see Supplementary Material). Therefore, for the sake of clarity in reporting, the diagram has been broken down to show behaviors and events around the following critical stages: Birth, Arrest, First murder, Trial. Alternative methods for data reduction in complex diagrams involve using more stringent cut-off criteria, such as only including transitions above a certain threshold frequency or standardized residuals above a value (see Ellis et al., 2017). However, to show all connections around critical areas, every transition was included³ around certain key times in serial killers’ life histories. Readers can see how these times fit into the whole life history by referring to the supplementary material.

State Transition Diagrams

The state transition diagrams (see Figures 1-4) present chains of behaviors. As the BSA conducted in the current study is a lag-one sequence analysis, only transitions between pairs of behaviors have been analyzed (e.g., A →; B, B →; C), which indicates B follows A more often than would be expected by chance, and C follows B more than expected by chance. It may seem intuitive to read this as A →; B →; C; however, this would be incorrect. In actual behavior terms in the current dataset, for example, “born” is an antecedent of “working class family,” and a sequitur of “Father left.” This means that a lot of serial killers were born into a working-class family. It also means that in a lot of cases, fathers left, and then the child was born. It does not mean there is a clear three-step progression of father left, then child is born, into a working-class family. This example also highlights the issue with causality in BSA. Obviously, the father leaving does not cause the child to be born, what the analyses show is that a father leaving was significantly more likely than chance to occur, before a child was born.

Figure 1.

State transition diagram of behaviors and events from birth.

Figure 2.

Behaviors and events surrounding first murder.

Figure 3.

Behaviors and events around being arrested.

Figure 4.

Behaviors and events during and following trial.

Sequences Related to Birth and Childhood

The first state transition diagram (see Figure 1) highlights the behaviors and events that occur after a child, who would later become a serial killer, is born. First, it was noted in a small but significant number of cases (n = 2, SR = 7.58) that the father had abandoned the child and family before the child was born. Whereas the original focus of the research was meant to begin from birth, the father leaving was included in the final analyses for completion. After being born, the most likely childhood environment reported in the biographies was an abusive childhood (n = 5, SR = 10.87), sometimes this occurred as part of a disciplinarian, strict, or religious household (n = 2, SR = 9.02). Indeed, several behaviors are shown around low economic status and poor or abusive living conditions. Abuse in childhood was linked to sexual assault crimes (n = 2, SR = 4.81). Although there was alcoholism reported in family members, it was not significantly linked to subsequent behaviors.⁴ Finally, some biographies indicated that the serial killer was injured or sickly in their childhood and had to seek medical attention (n = 3, SR = 26.18); however, this was not systematically or significantly linked to any of the reported childhood environments. What this means is that there is no clear link between the highlighted childhood environments and being sickly (it is equally likely across the different environments).

Sequences Related to First Murder

The sequences surrounding the first murder are more complex (see Figure 2), highlighting the range of modus operandi present in the current dataset. Sequences of later murders can be seen in the complete dataset (Supplementary Material). There are, however, some chains of behaviors that appear to be more likely than chance to occur across the sample. Murderers were almost equally likely to enter the victim’s home first (n = 4, SR = 10.09) or lure victims to their home (n = 5, SR = 11.84), rather than using other locations. Rape occurred more than chance when the murderer entered into the victim’s home (n = 2, SR = 6.84), or when the victim had been forced to choose a location to go to with the murderer (n = 3, SR = 13.91). If the victim was lured to the murderer’s home, then sex may still occur afterward (n = 2, SR = 16.85); however, original reports did not clarify that this was “rape”—therefore, it is included as “sex.” Ligature strangulation occurred more as a cause of death after rape (n = 6, SR = 14.70), rather than after having sex with the victim in the victim’s home (n = 2, SR = 12.67). Methods of “disposing” of the body, postmortem, varied between simply dumping the bodies after strangulation (n = 2, SR = 10.28), or keeping them in the (killer’s) house (n = 2, SR = 7.87), to be buried later (n = 3, SR = 11.92). Dismemberment or destruction of the corpse followed stabbing or slitting of the body more often than chance (n = 2, SR = 10.28). Several murderers did keep souvenirs after burying or disposing of the bodies (n = 4, SR = 15.98). If murderers kept souvenirs of the victims, they were more likely to revisit the crime scene (n = 2, SR = 17.41).

Arrested

Some serial killers have a history of being arrested for crimes (see Figure 3) of burglary (n = 2, SR = 4.44), assault (n = 2, SR = 4.44), or other miscellaneous crimes (n = 2, SR = 5.24). However, many more serial killers have a history of also being arrested for theft (n = 9, SR = 11.88), and crimes of sexual assault (n = 10, SR = 10.23), which is most likely an indicator of the later sexual homicide behaviors in the sample. Outcomes after being arrested are less clear, which is most likely an issue of the information available for analysis. In some cases, it is documented that sometimes a confession was gained following arrest (n = 4, SR = 5.16); however, in more cases, a trial followed being arrested (n = 11, SR = 9.88). A lot of sources simply stated that the serial killer was arrested and went to prison (n = 11, SR = 13.48); this means that information was not available regarding the confession and trial process. It should be noted that after serving their sentence, several cases in the sample then went on to commit crimes of sexual assault (n = 3, SR = 5.76).

Trial

To understand the legal process, sequences of events were mapped through the trial process (see Figure 4). Following trial, the most likely sequitur outcome was to be found guilty (n = 14, SR = 19.53).⁵ Following a guilty verdict, slightly more serial killers were likely to receive life sentences (n = 10, SR = 21.93) than death sentences (n = 9, SR = 18.74). Insanity pleas were more likely than chance to be rejected (n = 2, SR = 21.37), as was a defense plea of the murder being a mercy killing (n = 2, SR = 21.37).

Discussion

The aim of the current research was to provide a novel approach to understanding life histories of convicted serial killers, to develop descriptive models based on serial killers’ life histories for future behaviors. Using BSA, the research outlines transitions between behaviors and events that occurred across a sample of 25 well-known serial killers’ lives. Results first show that certain behaviors and events frequently occur in serial killers’ lives, which complements existing findings in the research literature. However, rather than follow a traditional approach of highlighting risk factors, the current research took a step further and analyzed these factors in chronological order. The current findings, therefore, can be seen to support existing, known findings; however, the BSA approach may give investigators and researchers a step toward being able to describe temporal pathways of serial killers’ lives.

The finding that serial killers suffer abuse in their childhood is well-documented in the literature (Arndt et al., 2004; LaBrode, 2007; Miller, 2014; Reid, 2017). These findings suggest that while childhood abuse is more likely than chance to occur in the life histories’ of serial killers, it does not occur in all cases, and investigation and profiling of serial killers may be hindered by this expectation. Indeed, serial killers may claim abuse as a means of attempting to gain reduced sentences at trial, or post-trial justification of their behavior. Again, it is important that the reader understands the current research is not claiming that “abuse leads to serial killing,” far from it. But, the current research does support the finding that abuse is apparent in some serial killers’ lives, and that this factor is part of a temporal chain that leads to other behaviors and events, which in certain cases, leads to serial killing.

Traditionally, the predominant view in the research indicates that females were almost exclusively the victims of serial killers (Arndt et al., 2004; Harbort & Mokros, 2001). However, other research has suggested that homosexual serial killers may target males (Miller, 2014), and with the acknowledgment of female serial killers came the further addition of males and family members as potential victims, too. The current findings fit with the traditional view of victim selection, with African American women, prostitutes, and female victims being the most likely victims. Notably, the current sample did include male victims, young victims, elderly victims, and familial victims; however, despite this, females were still the most likely victims. This may be a reflection of the larger number of male serial killers and their victims in the current dataset; however, this imbalance reflects wider-world statistics on the number of male to female serial killers.

Also, while not all serial killers are sexually motivated, the occurrence of rape was present in nearly half of the current sample and adds support to the expectation of sexual elements in serial murder (Arndt et al., 2004; Harbort & Mokros, 2001). However, this also adds support to arguments that claim that involvement of a sexual element is not a prerequisite of serial killing. In addition, rather than simply noting rape occurs, it is the behaviors directly preceding and following the act of rape that become important for investigations and profiling in terms of linking crimes together, and seeing changes in modus operandi. The current sample included fewer female serial killers, which reflects the base rates in known cases; however, future research should investigate the similarities and differences between male and female serial killers.

Dealing with changes in modus operandi is an issue for some traditional approaches to research and linking crimes. Indeed, the temporal nature of criminals’ activities changing and adapting to inter- and intrapersonal factors means that analyses should include temporal processes, so that change can also be analyzed. BSA allows for such changes, in terms of whether modus operandi changed or remained consistent. This is important as it allows an understanding of which criminals might change their style or methods of killing, and which remain constant. Retrospectively, this may mean that unsolved murders could be reviewed as victims of known offenders who have changed modus operandi. Prospectively, this could also be used to allow investigators the opportunity to understand which criminals may change their methods as they continue. This could, eventually, be used as a predictive approach to crime investigation; however, much more (detailed) data would be required before BSA models can be used in this way. The predictive process could be used by seeing how closely an ongoing/current case matches existing, known sequences. If a close match is found, then previous cases could be used to forecast future events and behaviors in serial killers’ actions.

Previous research has also indicated that serial killers may have an extensive criminal history (Miller, 2014), and that serial killing may be a result of an escalation of other crimes (Arndt et al., 2004). Consistent with previous research is the finding that being arrested was the most frequent individual event occurring in the current sample. Specifically, crimes of sexual assault and crimes of theft were the antecedent behaviors most likely to lead to individuals being arrested. This may contribute to the literature regarding risk factors for serial killing, and although not all individuals who commit crimes of theft or sexual assault will later become serial killers, a history of arrests that includes these crimes may indicate an individual more likely to be a suspect. BSA may, therefore, be used to narrow down a list of suspects, or allow police to rank the order of likelihood of suspect.

The current research is a first step toward showing how BSA could be used to map serial killers’ life histories; therefore, the sample size was rather small, which may limit the generalizability of the research to all serial killers’ behaviors. However, despite the limited sample size, the serial killers included are representative of gender, multiple nationalities, number of victims, and locations. In addition, significant transitions were still found in the dataset, which indicates that even with few cases, significant patterns and pathways in these data can be observed.

Sample size is a difficult issue in BSA research. The method can be used on a sample of 1, showing repeated behaviors across a life history of episode. Similarly, the method can be used on big data, of thousands of individuals. Therefore, there is no clear number needed to run the analysis. Instead, readers and researchers should remain aware of the size of the sample used, and the strength of relationships between events and behaviors. A strength of the research is that it is “additive”—newer cases can be coded and added to the existing dataset in a complementary way. This means that results can be updated as new data become available, which would show which transitions become stronger, and which diminish.

The current research has direct practical value to legal settings, in which large datasets may not be available or viable for coding. Future research, however, should seek to extend the current study by including more serial killers, with a range of backgrounds and methods of murder. This will then allow future investigators to highlight particular pathways pertinent to their investigation. For example, if the investigation sees a pattern of crime scene behaviors, these can be highlighted within a large dataset, and other behaviors and events predicted. Another area that the current research could be expanded upon is to include geographical information into the dataset. For instance, knowing the location of crimes (abduction, murder, and disposal of bodies) would allow a sequential map to be developed. This method has been suggested previously in the literature (Keatley, 2016, 2018); however, it was not possible with the information in the current dataset.

Conclusion

The current research used BSA to show the temporal relations of risk factors in serial killers’ life histories. Rather than focusing on individual frequencies of risk factors, BSA looks at the transitions between events and behaviors. These findings provide maps, or state transition diagrams, which can be easily read and interpreted by a wider audience of investigative and legal backgrounds. The current findings are consistent with previous findings, thus validating BSA as an approach to studying serial killing, and have identified new areas of interest. Key areas of interest were outlined in terms of upbringing environments, crimes leading to arrest, and methods of murder.

Supplemental Material

Supplementary_Material – Supplemental material for Using Behavior Sequence Analysis to Map Serial Killers’ Life Histories

Supplemental material, Supplementary_Material for Using Behavior Sequence Analysis to Map Serial Killers’ Life Histories by David A. Keatley, Hayley Golightly, Rebecca Shephard, Enzo Yaksic and Sasha Reid in Journal of Interpersonal Violence

Footnotes

Authors’ Note

David A. Keatley is now a senior lecturer in Criminology at School of Law, Murdoch University. Enzo Yaksic is currently affiliated with the Murder Accountability Project, VA, USA, the Serial Homicide Expertise and Information Sharing Collaborative, MA, USA and Researchers in Behaviour Sequence Analysis (ReBSA).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

Supplementary material for this article is available online.

Notes

Author Biographies

David A. Keatley is a senior lecturer in forensic psychology at the University of Lincoln, United Kingdom. He specializes in the areas of impulsive behaviors and sequence analysis. He is the director of Researchers in Behavior Sequence Analysis (ReBSA). He started his career at the University of Nottingham, United Kingdom, on an Economic and Social Research Council (ESRC)-funded PhD studentship. He has worldwide collaborations and links with academics and stakeholders, as well as police and investigators. He has multiple research articles published in international, peer-reviewed journals, and has presented at many international conferences.

Hayley Golightly is a graduate from the University of Lincoln. She is a member of ReBSA and has expertise in sequence analysis and has applied forensic experience. She is involved in several other areas of forensics and BSA.

Rebecca Shephard is a graduate from the University of Lincoln. She is a member of ReBSA and has expertise in sequence analysis and has applied forensic experience. She is involved in several other areas of forensics and BSA.

Enzo Yaksic founded the Serial Homicide Expertise and Information Sharing Collaborative (SHEISC) after 10 years of researching serial homicide where he acts as the database coordinator. As a contributor and reviewer for the Radford/FGCU Serial Killer Database Research Project, he worked to help establish the largest repository for information on atypical homicide offenders. His current work with the Murder Accountability Project aims to educate Americans on the importance of accurately accounting for unsolved homicides within the United States by using the Supplementary Homicide Report to help uncover suspicious clusters of potential serial homicide activity.

Sasha Reid is a third-year PhD candidate in Applied Psychology and Human Development at the University of Toronto. She is also a member of the Collaborative Doctoral Program in Human Development at the Fraser Mustard Institute. Her dissertation focuses on the developmental antecedents, biological contributions, and intrapsychic processes/worldviews of serial killers. Through her publications, she presents a more complete picture of serial homicide, its etiology, and the underlying motivations that exacerbate this unique form of fatal violence. At the intersection of developmental psychopathology, phenomenology, and biogenic criminology, this project is the first to explore how multiple developmental factors unfold and intersect to explain serial homicide.

References

Allely

C. S.

Minnis

Thompson

Wilson

Gillberg

(2014). Neurodevelopmental and psychosocial risk factors in serial killers and mass murderers. Aggression and Violent Behavior, 19, 288-301. doi:10.1016/j.avb.2014.04.004

Arndt

W. B.

Hietpas

Kim

(2004). Critical characteristics of male serial murderers. American Journal of Criminal Justice, 29, 117-131. doi:10.1007/BF02885707

Bakeman

Gottman

J. M.

(1986). Observing behavior: An introduction to sequential analysis. Cambridge, UK: Cambridge University. doi:10.1017/CBO9780511527685

Canter

Missen

Hodge

(1996). A case for special agents. Policing Today, 2(1), 2-11.

Clarke

D. D.

Crossland

(1985). Action systems: An introduction to the analysis of complex behaviour. London, England: Methuen.

Douglas

Burgess

A. W.

Burgess

A. G.

Ressler

R. K.

(1992). Crime classification manual: A standard system for investigating and classifying violent crimes. Chicago, IL: John Wiley & Sons.

Ellis

H. E.

Clarke

D. D.

Keatley

D. A.

(2017). Perceptions of behaviours in stranger rape cases: A sequence analysis approach. Journal of Sexual Aggression, 23, 328-337. doi:10.1080/13552600.2017.1361618

FBI. (2008). Serial Murder: multi-disciplinary perspectives for investigators. Behavioural Analysis Unit: National Centre for the Analysis of Violent Crime

Fossi

J. J.

Clarke

D. D.

Lawrence

(2005). Bedroom rape: Sequences of sexual behavior in stranger assaults. Journal of Interpersonal Violence, 20, 1444-1466. doi:10.1177/0886260505278716

10.

Harbort

Mokros

(2001). Serial murderers in Germany from 1945 to 1995: A descriptive study. Homicide Studies, 5, 311-334. doi:10.1177/1088767901005004005

11.

Hickey

E. W.

(1992). Serial murderers and their victims. Contemporary Sociology, 21, 373-374. doi:10.2307/2076297

12.

Hodgkinson

Prins

Stuart-Bennett

(2017). Monsters, madmen . . . and myths: A critical review of the serial killing literature. Aggression and Violent Behavior, 34, 282-289. doi:10.1016/j.avb.2016.11.006

13.

Ioana

I. M.

(2013). No one is born a serial killer! Procedia–Social and Behavioral Sciences, 81, 324-328. doi:10.1016/j.sbspro.2013.06.436

14.

Ivanouw

(2007). Sequence analysis as a method for psychological research. Nordic Psychology, 59, 251-267. doi:10.1027/1901-2276.59.3.251

15.

Keatley

D. A.

(2016). Understanding patterns of action: Sequence analysis in military contexts. Oral Presentation in: 5th Military Psychology Conference; November 2, 2016, The Ark Conference Centre, Basingstoke.

16.

Keatley

D. A.

(2018). Pathways in crime: An introduction to behaviour sequence analysis. London, England: Palgrave Macmillan.

17.

Keatley

D. A.

Barsky

A. D.

Clarke

D. D.

(2017). Driving under the influence of alcohol: A sequence analysis approach. Psychology, Crime & Law, 23, 135-146. doi:10.1080/1068316X.2016.1228933

18.

LaBrode

R. T.

(2007). Etiology of the psychopathic serial killer: An analysis of antisocial personality disorder, psychopathy, and serial killer personality and crime scene characteristics. Brief Treatment and Crisis Intervention, 7, 151-160. doi:10.1093/brief-treatment/mhm004

19.

Lawrence

Fossi

Clarke

(2010). A sequential examination of offenders’ verbal strategies during stranger rapes: The influence of location. Psychology, Crime & Law, 16, 381-400. doi:10.1080/10683160902754964

20.

Leitner

Kent

(2009). Bayesian journey-to-crime modelling of single and multiple crime-type series in Baltimore County, MD. Journal of Investigative Psychology and Offender Profiling, 6, 213-236. doi:10.1002/jip.109

21.

Marono

Clarke

D. D.

Navarro

Keatley

D. A.

(2017). A behaviour sequence analysis of nonverbal communication and deceit in different personality clusters. Psychiatry, Psychology and Law, 24, 730-744. doi:10.1080/13218719.2017.1308783

22.

Marono

Clarke

D. D.

Navarro

Keatley

D. A.

(2018). A sequence analysis of nonverbal behavior and deception. Journal of Police and Criminal Psychology, 33(2), 109-117.

23.

Miller

(2014). Serial killers: I. Subtypes, patterns, and motives. Aggression and Violent Behavior, 19, 1-11. doi:10.1016/j.avb.2013.11.002

24.

Mott

N. L.

(1999). Serial murder: Patterns in unsolved cases. Homicide Studies, 3, 241-255. doi:10.1177/1088767999003003003

25.

Reid

(2017). Compulsive criminal homicide: A new nosology for serial murder. Aggression and Violent Behavior, 34, 290-301. doi:10.1016/j.avb.2016.11.005

26.

Ressler

R. K.

Burgess

A. W.

Douglas

(1988). Sexual homicide: Patterns and motives. New York, NY: Simon & Schuster.

27.

Simkin

M. V.

Roychowdhury

V. P.

(2014). Stochastic modeling of a serial killer. Journal of Theoretical Biology, 355, 111-116. doi:10.1016/j.jtbi.2014.03.039

28.

Stone

M. H.

(2001). Serial sexual homicide: Biological, psychological, and sociological aspects. Journal of Personality Disorders, 15(1), 1-18. doi:10.1521/pedi.15.1.1.18646

29.

Taylor

Keatley

D. A.

Clarke

D. D.

(2020). A behavior sequence analysis of perceptions of alcohol-related violence surrounding drinking establishments. Journal of Interpersonal Violence, 35(9-10), 1982-1997.

30.

Taylor

P. J.

Jacques

Giebels

Levine

Best

Winter

Rossi

(2008). Analysing forensic processes: Taking time into account. Issues in Forensic Psychology, 8, 43-56.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.11 MB