Serial Domestic Abuse in Wales: An Exploratory Study Into its Definition, Prevalence, Correlates, and Management

ABSTRACT

Analysis of information held by police, probation, and third-sector organizations in Wales about 100 domestic abuse perpetrators, along with 16 practitioner interviews, provides the empirical context for a discussion of the problem of “serial domestic abuse.” Despite increased concern over the harm caused by serial abusers, different definitions and recording systems prevent a reliable estimation of the problem. This exploratory study suggests that the offending profiles of serial abusers are heterogeneous, and recommends that approaches aimed at reducing the harm caused by the “power few” domestic abusers incorporate information about serial alongside repeat and high-risk offending.

KEYWORDS:

• domestic abuse
• harm
• high risk
• perpetrators
• repeat offending
• serial offending
• Wales

Introduction

In its various forms, domestic abuse is a serious problem and, in whatever guise it takes, is rarely a one-off occurrence (European Union Agency for Fundamental Rights [FRA], 2014). Domestic abuse has a higher rate of repeat victimization than any other crime, accounting for 76% of all incidents (Smith, Flatley, & Coleman, 2010). More so than any other crime type, victims of domestic abuse report the lowest levels of personal well-being and life satisfaction (Office of National Statistics [ONS], 2015). In the past decade, the development of a more skilled and professionalized range of services has led to a more nuanced understanding of domestic abuse and its various manifestations, including coercive control, stalking, intimate partner rape, and most recently serial domestic abuse (that which a single perpetrator commits against multiple victims).

Existing scholarship has revealed that the repetitive nature of domestic abuse is one of its key distinguishing features, central to its conceptualization as “a coercive course of conduct, usually involving a series of related occurrences, rather than a one-off event” (Walby, 2005, p. 4; see also Stark, 2007). Research demonstrates that the majority of male domestic abuse perpetrators are repeat offenders, with English research producing a figure of 83% within a six-year period (Hester, 2013) and American research finding 60% within a ten-year period, although this was deemed to be an underestimate because recidivism was measured as new arrests rather than incidents (Klein & Tobin, 2008). Unlike other crime types, for domestic abuse there are “high frequency victims,” and accurately capturing their experiences poses challenges for measurement and analysis of national crime victimization survey data (Farrell & Pease, 2007; Walby, 2015; Walby, Towers, & Francis, 2014). The detrimental consequences for victims and their children of repeated exposure to domestic abuse has been recognized in the widespread adoption in the United Kingdom of responses aimed at focusing resources on those victims at highest risk of re-abuse (e.g., multiagency risk assessment conferences and independent domestic violence advisors; see Robinson, 2010, for an overview). Although repeat offending against the same partner is a fundamental aspect of research, theory, and policy responses to domestic abuse, less is known about the ways in which serial perpetrators abuse multiple partners.

The idea of the “serial domestic abuser” has emerged in recent years to typify the most dangerous type of domestic abuser. This feeds into our fearful expectations of serial criminals as responsible for the most “severe criminal and harmful behaviors in society” (Edelstein, 2016, p. 62). Not surprisingly, when domestic abuse is made visible by the British media, it is often the most serious and disturbing cases, as is committed by some serial abusers.¹ Perhaps the most well-known example from the United Kingdom is the murder of Clare Wood by the serial abuser George Appleton, which took place in England in February 2009.² Statements in the popular press by senior criminal justice officials have reinforced the view that serial abuse is an especially problematic and harmful form of domestic abuse. “We want to go after those serial perpetrators who go from relationship to relationship, growing in confidence and menace,” said Chief Constable Brian Moore following publication of a high-profile report by the U.K.’s Association of Chief Police Officers (ACPO) that focused on responding to domestic abuse perpetrators (Association of Chief Police Officers, 2009). The same report contended that serial domestic abuse constitutes a sizeable problem in the United Kingdom, with an estimated 25,000 serial abusers in contact with the police at any one time. Despite an increased level of concern over the harm posed by these offenders, however, an evidence-based understanding of the problem to underpin recommendations for appropriate policy responses remains lacking. The overall aim of the present study is to achieve a clearer picture of how serial domestic abuse is being understood, defined, and recorded by criminal justice and partner agencies in Wales. A specific focus on serial domestic abuse provides much needed information on the definition, prevalence, correlates, and possibilities for the effective management of this “new” social problem.

Previous research into serial domestic abuse

Roughly 40 years of interdisciplinary scholarship has produced an extensive body of knowledge in relation to domestic abuse (Stark & Buzawa, 2009). For example, the recent review by the U.K.’s National Institute of Health and Care Excellence (NICE) identified more than 50,000 English-language domestic abuse references for the period 2000–2012 alone (National Institute of Health and Care Excellence, 2014). Their wide-ranging and systematic review of the literature found that research pertaining to perpetrators and effective interventions for dealing with perpetrators was relatively limited, and no specific recommendation relating to perpetrators was made other than to “commission and evaluate tailored interventions for people who perpetrate domestic abuse” (i.e., conduct further research). Against this backdrop, perhaps it is to be expected that few studies are available that specifically examine serial domestic abuse.

Definition

Exactly what is meant by the term “serial domestic abuse” is ambiguous, with definitions used by researchers and practitioners alike varying considerably. Definitions may vary according to the timeframe imposed (e.g., within the past year, past three years, or the perpetrator’s entire known criminal history) as well as the number of victims (e.g., two or three or more victims) and type of criminal offenses included (e.g., intimate partner violence only or violence against women more broadly defined). This is not a new problem, nor is it confined to this particular topic. Prior to 1980 serial crimes were not a differentiated type, and since then most research on serial criminality has focused on murder, rape, and arson (Kocsis & Irwin, 1998). Conceptual and empirical definitions have long been contested among researchers studying the phenomena of criminal careers and/or serial criminality, a problem compounded by the passage of time (Edelstein, 2016). Directly pertinent to the current study, in 2009 the ACPO recommended a definition of serial domestic abuse (“the perpetrator is alleged to have used or threatened violence against two or more victims who are unconnected to each other and who are or were intimate partners of the perpetrator, noting the important distinction from repeat offending against the same victim or persons in the same household” (p. 19, emphasis added)); however, it is unclear to what extent this policy definition has penetrated the practice of British police or other relevant agencies. At the time of writing, the U.K. College of Policing recently published the following definition for use by relevant agencies: “A serial perpetrator is someone who has been reported to the police as having committed or threatened domestic abuse against two or more victims. This includes current or former intimate partners and family members” (College of Policing, 2015). The assumption underpinning these efforts is that a standardized definition will help facilitate the establishment of a more systematic process for routinely identifying and managing serial abusers, thereby contributing to a more effective overall response.

Prevalence

Richards’s (2004) strategic overview of approximately 400 serious domestic abuse and domestic sexual assault cases in the London area found evidence of significant reoffending, including an (unspecified) number of men who are “serial offenders, who go from one abusive relationship to the next, are violent to significant women as well as other people in their lives” (p. 8). This research, while important for documenting the serious and prolific offending of some domestic abuse perpetrators, unfortunately conflated serial domestic abuse with other types of serial violent offending. A few years later, Hester and Westmarland (2007) were able to produce a specific prevalence estimate for serial domestic abuse using anonymized data from one police force area in England. Their profile of 692 perpetrators reported to the police for domestic abuse found that 50% were involved in an additional incident during the three-year follow-up period, and 18% of those that reoffended did so against a subsequent new partner. This is remarkably similar to the figure of 17% in another English police force area (Bland & Ariel, 2015). American research, using perpetrator samples derived from criminal justice rather than police data, has resulted in slightly higher prevalence rates of serial domestic abuse, ranging from 28% of probationers within a one-year period (Klein, Wilson, Crowe, & DeMichele, 2005) to 43% of persons arrested for violating a civil restraining order over a six-year period (Bocko, Cicchetti, Lempicki, & Powell, 2004). In sum, existing research indicates that serial abusers constitute a sizeable minority of the total number of domestic abusers known to criminal justice agencies at any one time, but these estimates vary according to the data source and the definition being used.

Correlates

Given the dearth of research in this area, few studies are able to demonstrate whether there are distinctive correlates associated with serial domestic abuse. Research on other serial violent offenders suggests that what distinguishes them from nonserial offenders is a “particular type of propensity to re-offend” based on psychological factors (Kocsis & Irwin, 1998, p. 201), such as impulsivity (Baltieri & De Andrade, 2007). However, extant research on domestic abuse has found that abusers who go on to abuse new partners are not substantially different from those who reabuse the same partner (i.e., the profiles of serial and repeat abusers are more alike than different). There is some evidence that serial abusers tend to be younger and less likely to be married (Adams, 1999; Klein et al., 2005) but, perhaps contrary to popular opinion, extant research does not indicate that serial abusers are the most dangerous, particularly when they are compared to repeat abusers. More detailed research, especially studies designed to capture information about victim sequelae, would be beneficial for establishing whether serial domestic abusers are a distinctive group and indeed whether the nature of their offending is experienced by victims as more harmful.

Management

A desire for more proactive and targeted approaches to manage the risk posed from the perpetrators of domestic abuse, most notably serial perpetrators, has been highlighted in a range of U.K. policy documents. Such efforts are underpinned by Sherman’s (2007) concept of the “power few” (i.e., a small percentage of offenders is responsible for a high percentage of harm) and expectations that investing resources in identifying this group, and reducing their offending, will pay dividends in terms of harm reduction. For example, one of the recommendations in the 2009 ACPO review provided the initial call for “wider recognition, and improved management, of serial perpetrators of violence against women and girls.” Five years later, the 2014 inspection by Her Majesty’s Inspectorate of the Constabulary (HMIC, 2014) called for “examples of how forces are targeting serial and repeat domestic abuse perpetrators in order to prevent future offending” (p. 24), which led to the implementation of a national working group, with membership from several English and Welsh agencies, with responsibility for taking this area of work forward (Chamberlain, 2014). However, a recent follow-up investigation by HMIC (2015) found that there remains a “lack of consistency around the management of serial perpetrators” (p. 97) and there are presently few British examples of responses aimed specifically at serial abusers. Those that have been implemented include enhanced monitoring alongside a range of preventative policing actions (e.g., informing neighborhood policing teams of the perpetrator’s history; flagging addresses, so any calls to the police are prioritized; undertaking home visits; surveillance; and working with partner agencies to address problems such as substance abuse) (Labor Party, 2012). The extent to which empirical research has informed these approaches, and whether they produce the desired outcomes for the victims and perpetrators of serial domestic abuse, remains unknown.

Exploratory research in Wales

The Welsh context

At the time this research was being conceived, there was a heightened anxiety around domestic abuse perpetrators, particularly the “power few” considered responsible for a disproportionate number of crimes. There was increased attention being paid to serial perpetrators, both nationally (as previously mentioned) and in Wales. For example, the triple murder by Carl Mills of three generations (grandmother, mother, infant girl) of one Welsh family occurred in September 2012 and emerging revelations about missed opportunities to intervene were reverberating through the practitioner community in the following months. It was felt that existing responses to domestic abuse perpetrators could be more proactive, and more effective.

Additionally, the structure and remit of probation services in Wales changed during the course of this research. From June 1, 2014, the National Probation Service (NPS) and 21 Community Rehabilitation Companies (CRCs) were formed to replace the 35 former Probation Trusts. The NPS is a statutory criminal justice service that supervises high-risk offenders released into the community. The 21 CRCs have responsibility for managing low- and medium-risk offenders. This organizational restructuring was viewed at the time with trepidation by key stakeholders involved in commissioning the research, as it was seen to pose an additional barrier in the effective management of high-risk domestic abuse perpetrators, who might not always be assessed as such due to the well-known underreporting of incidents and the high attrition rate of cases (Hester, 2006). All of these factors combined served to reinforce a desire to gain more information about the nature of serial domestic abuse as a first step to considering how strategies focused on identifying and responding to serial abusers might contribute to an improved response to domestic abuse generally, which lead to the present study.

Methodological approach

Informed by this empirical and policy context, an exploratory study adopting a mixed-methods approach took place from December 2013 to October 2014 to provide some preliminary answers to the following research questions: How is serial domestic abuse being defined and understood by criminal justice and other relevant practitioners? Using current definitions, what is the prevalence of serial domestic abuse? What are the correlates of serial domestic abuse (i.e., can distinguishing features be identified and to what extent does serial offending overlap with repeat and/or high-risk offending behavior)? What are the implications of this research for informing agency responses designed to manage perpetrators and reduce domestic abuse? The overall objective of the research was to provide some new research evidence about the level and nature of serial offending from domestic abuse perpetrators in order to inform policy and practice initiatives being considered in Wales and elsewhere in the United Kingdom. The research involved the analysis of both qualitative and quantitative Welsh data, the specific details of which are described below.

Qualitative data

First, a qualitative mapping exercise was undertaken to provide a descriptive overview of both the domestic abuse recording processes and possible methods of identifying serial abuse perpetrators across different agencies in Wales. To facilitate this, a total of 16 semistructured interviews were conducted with a range of practitioners working in different agencies across Wales. The style and length of the interview depended on the practitioner’s role and agency. For example, two interviews were conducted while sitting alongside practitioners as they talked through their data recording systems. Some interviews were conducted over the phone due to geographical distance. The majority, however, were face-to-face interviews that lasted approximately one hour each and covered a range of topics associated with how their agency responds to domestic abuse.

Sample

The interview sample was achieved by snowball methods and included relevant individuals from Welsh police forces (n = 7) who were identified by contacting the Head of the Public Protection Team in each force. Two representatives from probation were interviewed,³ along with a number (n = 7) of practitioners working in health and third-sector agencies with a focus on domestic abuse (five worked in domestic abuse organizations providing services to victims, one held a strategic role for domestic abuse within local government, and one worked in a health team providing additional care for women with complex needs). Interviews were conducted by a research assistant with the explicit consent of participants, and covered some or all of the following topics: (1) definitional issues around serial domestic abuse, (2) identification and flagging of cases, (3) recording systems and processes, (4) deployment of resources and operational issues, (5) legislative context, (6) information sharing with partner agencies, and (7) use of screening tools to assess perpetrators of abuse. The research assistant, an experienced qualitative researcher, conducted open, axial coding to analyze the data (Strauss, 1987). For the purposes of this paper, the main themes discussed include those relating to the definition of serial domestic abuse, and practitioners’ perspectives on how these individuals should be managed by criminal justice and partner agencies. Where direct quotes are used, information about the respondent’s role and/or agency has not been provided in order to preserve their anonymity.

Quantitative data

Given the variability of recording and monitoring practices revealed by the qualitative mapping exercise, the research could not draw upon existing case-file recording systems for information, due to the absence of a dedicated question, risk factor, or “flag” that could be utilized to identify serial abusers. Therefore a bespoke database was created to assist practitioners to retrieve information from their agency’s databases about a random sample of 100 domestic abuse perpetrators (see next section). The database took the form of an Excel spreadsheet that included columns to describe (1) the nature of the offending (e.g., from police incident, arrest and/or intelligence data, probation risk assessment data, etc.); (2) the victims involved (e.g., number, relationship, and other details); (3) timeframe (e.g., within last year, three years, and/or lifetime); and (4) the practitioner’s assessment as to whether the information available from their agency indicated serial offending. Information and guidance about each column was embedded within the spreadsheet, and participating practitioners (representing police, probation, and third-sector organizations) were provided with the contact details of the research team in the event they needed to ask for any further clarification. This exercise enabled each agency to identify the proportion of the random sample that they deemed to be serial abusers, as well as revealing the gaps and overlaps in the information held by different agencies about particular perpetrators.

Added to this database was other detailed information about the perpetrators, held by Wales Probation Trust, that could be used to establish the characteristics of those identified as having committed serial offending, compared with those who were not identified as such. At this point it will be useful to provide a few more details about the nature of the data routinely collected by probation officers in Wales. First, an OASys⁴ assessment is completed for all offenders supervised by probation and is used to assess an offender’s risk of reconviction, the factors associated with offending, and the risk of harm the offender presents. The results of the OASys are used to determine the level of intervention and supervision to be put in place to reduce risk of reoffending. Offenders are assessed prior to sentencing, at the start of their community or custodial sentences, and reviewed at several points during the sentences. The OASys includes information related to five areas: (1) risk of reconviction and offending-related factors; (2) risk of serious harm to others, risks to the individual, and other risks; (3) the OASys summary sheet; (4) the sentence plan; and (5) self-assessment. The first consists of 13 sections covering a range of offending-related issues, one of which focuses upon relationships and past or current domestic abuse. Upon disclosure of domestic abuse by an offender, probation officers are meant to complete the Spousal Assault Risk Assessment (SARA), which consists of 20 risk factors along with two summary risk judgments (Kropp, Hart, Webster, & Eaves, 1995); however, it is clear from the data that probation officers do not comply with this requirement in every case (see Table 1). The SARA is meant to be completed regardless of the current marital or relationship status of the offender. Finally, other available information added to the database included geographic region of Wales; sex; race/ethnicity; age at time of assessment; age at first conviction; parenting responsibilities; alcohol, drugs, mental health issues; SARA risk factors; and summary risk judgments (OASys risk of harm, SARA risk to partner, SARA risk to others).

Table 1. Comparison of random sample (n = 100) and population (n = 6,642) of Welsh domestic abuse perpetrators.

Variable	Sample	Population	P-value
Region of Wales
Northwest	10.1%	13.7%	.076
Northeast	18.2%	9.9%
Central	18.2%	13.0%
West	6.1%	8.7%
Southwest	16.2%	16.7%
South	17.2%	21.1%
Southeast	14.1%	16.9%
Sex (male)	96.0%	94.1%	.525
Ethnicity (white British)	96.8%	95.5%	.823
Age at time of assessment (mean years)	35.4	33.9	.143
Age at first conviction (mean years)	21.3	19.8	.084
Parenting responsibilities (yes)	63.9%	52.7%	.017
Alcohol misuse (yes)	68.0%	64.4%	.528
Drug misuse (yes)	43.0%	42.0%	.839
Mental health problems (yes)	47.9%	48.1%	.527
OASys Risk of Harm
Low	7.1%	4.5%	.507
Medium	16.5%	78.0%
High	15.3%	17.0%
Very high	1.0%	0.5%
SARA Risk to Partner
Low	13.6%	12.9%	.984
Moderate	72.7%	73.0%
High	13.6%	14.1%
SARA Risk to Others
Low	39.4%	33.7%	.597
Moderate	53.0%	56.9%
High	7.6%	9.4%

Note: Depending on the measurement of variables, chi-square or F-values were calculated. Across all variables, missing data was less than 5%, except for ethnicity (7%) and the SARA variables (34%).

Sample

The bespoke database described above was completed for a random sample of 100 perpetrators. This sample was drawn from an anonymized dataset provided by Wales Probation Trust representing the population of domestic abuse perpetrators known to the agency for the 2013–2014 financial year.⁵ Specifically, the dataset contained a total of n = 6,642 individual domestic abuse perpetrators, representing all perpetrators who had their first assessment with a probation officer completed in Wales during the period.⁶ Although official data from Wales Probation Trust could not be used to answer the research questions for this study as it did not contain a variable about serial offending, it did provide a rich, national data source from which to sample a smaller group for further study, using the bespoke database. Table 1 provides a comparison of the population of perpetrators and the random sample on a range of demographic and offending variables, and demonstrates that they are statistically comparable on all measures except for parenting responsibilities.

Research findings

Lack of consistent definition

The research established that, whilst all Welsh police forces have a definition of serial domestic abuse in place, each varies slightly. Two of the four forces in Wales were using a definition consistent with that recommended by ACPO in 2009 (i.e., two or more unconnected victims, over an unspecified length of time). The other two forces made their definitions more stringent, either by increasing the minimum number of victims or specifying a time limit. For example, the third Welsh force required a threshold of three victims and the fourth imposed a 12-month time period. Police respondents felt there was a need for a standardized definition and agreed that the current inconsistency of approach could lead to an individual being labelled a serial abuser in one force area and not another. Further, while all forces utilized a definition, the extent to which this definition was made explicit to officers working outside of specialized domestic abuse teams was not clear. Although all forces reported that they had issued guidance to local policing units regarding the recently revised U.K. governmental definition of domestic abuse,⁷ forces’ definitions of serial domestic abuse were not necessarily included within this guidance. In some areas, only officers with specialist knowledge were aware of how their force was defining serial abuse:

You would struggle to find a definition for serial abuse written down anywhere. It’s those who need to know, know. The distinction between a serial and repeat perpetrator … you won’t find that written down anywhere, it’s inherited knowledge… . I think we have to be honest with ourselves and say we need to rubber stamp our definition somewhere and circulate it as people can’t be asked to act on something if they don’t know what criteria they are applying to someone.

It was acknowledged that the level of ambiguity in defining the problem could only have negative implications for the effectiveness of responses developed to address it, regardless of what these might be.

The lack of a consistent definition for serial domestic abuse did not appear to be an issue for police forces alone. Other agencies (including probation, health, social services, and those in the third sector such as domestic abuse organizations supporting victims) did not have any definitions in place, although practitioners working in these agencies were under the impression that serial domestic abuse was a problem. For example, although offender assessments carried out by probation incorporate previous domestic abuse and relationship history into the risk of harm analysis, the two probation staff interviewed were not aware of one standardized definition for a serial perpetrator in place within probation. Indeed, interview respondents referred to “their own” definition and were aware of other staff working to different definitions of serial domestic abuse. All of the staff interviewed from health and third-sector agencies indicated that they were unaware of any standardized definition of serial abuse within their respective agencies, and expressed the view that a single agreed definition of serial domestic abuse would be preferable to the current situation.

No we don’t have such a definition. We just use local knowledge, say that he has been abusive and come to the attention of other agencies. That’s how we know our serial perpetrators. One definition would be so helpful, we need consistency, to look at their history and number of partners …

Although respondents were supportive in principle of the implementation of a standardized definition, through the course of the interviews many also conveyed the negative implications associated with the use of any definition, and who would be included and excluded. For example, the recent revision to the U.K. government definition of domestic abuse to include young people aged 16 and 17 was seen to invariably lead to a greater proportion of young offenders being labelled as domestic abuse perpetrators—and as a function of this, serial abusers. A few respondents expressed concern with labelling young people in this way, as illustrated by the following quotes:

The problem with a broad definition is that different kinds of people get caught up in it. Sixteen-year-olds tend to have more short-term relationships and they are more likely to live at home with family, in which case there are more likely to be two victims. Say a teenager falls out with his mum and his sister? There is a danger of stigmatizing and over labelling young people as serial offenders.

It’s something to be conscious of because they are children, and do you want to give a label to a child as being a serial domestic abuser? There could be a lot of other factors linked to the behavior.

Similarly, a few felt that the recommendation from ACPO to use a threshold of two or more victims to define a serial perpetrator was too low, and should be higher in order to avoid labelling a large proportion of offenders as serial abusers. Also, a more stringent definition could more effectively focus resources upon the highest risk and most dangerous minority.

Two victims, well a lot of people fall into that category. Two doesn’t really point to anything, that can’t make a pattern of behavior. The vulnerability of the victim, severity and risk of the serial perpetrators should be factored in.

Conversely, many respondents also raised concerns over definitions that would improperly exclude certain individuals. Individuals remaining in long-term relationships, for example, who then separate and subsequently go on to abuse further victims may not be defined as serial abusers through automated reporting/flagging processes as the entirety of their offending would not be included within the timeframes imposed by some definitions. A similar concern was expressed over those subject to lengthy custodial sentences. As one respondent commented:

I do recognize that perhaps someone may lapse for a while, could be because they have had a custodial sentence for five years and then they come out, start offending again and we don’t pick them up as serial perpetrators. But I suppose if they have had a substantial prison sentence and committed a serious crime, they are going to come out as a monitored person anyway …

From the perspective of frontline practitioners, although multiple definitions are a hindrance to effective multiagency responses, any single definition will also have its own set of limitations that must be understood and acknowledged.

Challenges establishing prevalence

In addition to the definitional issue noted above, several other issues pose a significant and substantial challenge to establishing reliable estimates of the prevalence of serial domestic abuse. These relate to problems associated with researching domestic abuse generally, which can be summarized as follows: (1) conceptualizations of domestic abuse have broadened over time, and now in most countries this term encompasses a range of behaviors beyond physical assault (e.g., the revised U.K. governmental definition now includes coercive control); (2) likewise, legal and policy definitions of what types of behavior constitute domestic abuse are dynamic, thus challenging attempts to reliably establish trends over time; (3) a lack of a specific criminal offense for domestic abuse in the United Kingdom and other jurisdictions makes tracking these types of cases through the criminal justice system more difficult; (4) the widespread and widely acknowledged underreporting of domestic abuse to police by victims and witnesses creates a substantial “dark figure”; (5) the historical and continued high attrition rates of domestic abuse cases through the criminal justice system indicate a significant “justice gap” in terms of offenders never sanctioned for their behavior; and (6) despite efforts to promote consistency and integration, variation in data and recording practices across agencies remains. Taken together, these issues strongly suggest that the picture any one agency has about the victims or perpetrators of domestic abuse must be considered, at best, incomplete. Establishing conclusively who is or who is not a domestic abuser—let alone whether their offending can be considered repeat, serial, and/or high risk—is indeed a very tenuous exercise. In short, the challenges to establishing reliable estimates of the size of the problem go well beyond the problem discussed in the previous section of establishing and implementing a common definition across agencies.

Through a detailed examination of a random sample of 100 perpetrator records, the present study revealed that the proportion deemed to be serial abusers varied considerably depending on the source of information and definition used (e.g., from police, probation, or third-sector agencies), ranging from 4% to 20%. The rich offender profiles obtained by probation officers during offender assessments offered enough information to determine that 13% were serial abusers (e.g., from data coded from the OASys, including the pattern and nature of the offending, when it took place, and who was involved). The data held by police forces (e.g., arrests, incidents, intelligence logs relating to domestic abuse offending for each perpetrator) produced an even larger estimate of 20%. Thus, the criminal justice–based percentage derived from the present study is broadly in line with the one other most comparable study—Hester and Westmarland’s (2007) estimate of 18%. However, it is worth noting the substantial variation in estimates across the four Welsh police forces. Keeping in mind the different sample sizes for each force, it is evident that some forces identify a much greater proportion of perpetrators as serial than do others (ranging from 0% to 27%). As expected, those operating with more stringent definitions produced lower estimates than those forces using the ACPO definition.

The information held by a small number of third-sector agencies participating in the research produced the lowest estimate (4%). As community-based victim service providers are seen to obtain disclosures of abuse that might not be reported to criminal justice agencies (Robinson & Payton, 2016), this finding was somewhat counterintuitive, although it could be partly explained by reference to the fact that these agencies typically hold information related to the victims rather than offenders. It should be acknowledged that only a small number of third sector agencies participated in this research and thus it is doubtful that the percentage obtained represents a reliable representation of the amount of information held in such agencies generally. Indeed, this study did find a few examples of agencies in the third sector knowing about additional victims that were unknown to other agencies and, in one case, highlighted domestic abuse that had not come to the attention of criminal justice agencies.

The challenge of making a robust determination about who is versus who is not a serial perpetrator was not only due to different agency information and recording systems. It was also complicated by files containing compelling evidence of other types of related offending (e.g., sexual violence or other forms of violence against women or girls). Although such information was not included in the percentages noted above, this is an issue worthy of discussion. For example, in five cases there was evidence that perpetrators committed violence against other family members (but there was no documented offending against more than one intimate partner), and in most of these the victims were female family members. Three perpetrators had evidence of serial sexual offending (in all cases unconnected victims who were underaged girls) but there was no known serial domestic abuse. Further research is needed to understand the implications of including or excluding such individuals from operational practices designed to identify and manage the most dangerous perpetrators.

Evaluating agency information and the gaps and overlaps across agencies revealed that different agencies were identifying different individuals as serial (from the same sample of 100 domestic abuse perpetrators) with only a very small degree of overlap across agencies. For example, in only 4 cases (“Trevor,” “Adam,” “Matt,” and “Malcolm”; see Table 2 below⁸) was the individual in question deemed to be a serial abuser by both police and probation. An explanation for why agency assessments were consistent for these 4 individuals in particular is not immediately apparent. For example, they varied in their ages and did not have a particular type of offending profile, nor were they deemed to be the more “risky” offenders, nor did they tend to be identified from one police force area. The very small degree of overlap is likely due to the different information about offending held by agencies; for example, relying on different sources (the offender him/herself versus victims and/or witnesses) at different stages (offending that is unreported, reported to police, results in arrest, conviction, etc.). The only commonality was that all 4 were positively identified as having committed domestic abuse within the past year, the past three years, as well as further back in their lifetimes, indicating perhaps the importance of a lengthy history of offending. On the whole, however, there seems to be more agreement among agencies about who is not rather than who is a serial abuser.

Table 2. Description of 13 serial domestic abusers from n = 100 random sample.

Pseudonym & demographic characteristics of perpetrator	Number and type of victims	Offending timeframe			Probation assessments
		DV in last year	DV in last 3 years	DV in lifetime	# OASys	# SARA
“Trevor” Male, 28, white British	Two offenses against two ex-partners	yes	yes	yes	2	0
“Adam” Male, 20, white British	Two offenses against two ex-partners	yes	yes	yes	2	2
“Mike” Male, 55, white other	Offenses against ex-wife and ex-partner	no	no	yes	3	3
“Matt” Male, 38, white British	Offenses against current partner and three previous partners	yes	yes	yes	10	5
“Chris” Male, 40, white British	Offenses against current wife and ex-wife	no	no	yes	11	5
“David” Male, 40, white British	Offenses against ex-wife and two ex-partners	no	yes	yes	17	12
“Ken” Male, 37, white British	Offenses against two ex-partners	no	yes	yes	1	4
“Fiona” Female, 31, white British	Offenses against two ex-partners	no	yes	yes	24	0
“Lawrence” Male, 45, white British	Offenses against two ex-partners	yes	yes	yes	3	2
“Harry” Male, 31, white British	Offenses against current partner and two previous partners	yes	yes	yes	27	7
“Sam” Male, 29, white British	Offenses against current partner and one previous partner	no	yes	yes	26	6
“Malcolm” Male, 30, white British	Offenses against two ex-partners	yes	yes	yes	15	1
“Nick” Male, 31, white British	History of offending against partners	unclear	unclear	yes	9	6

Note: All 4 female perpetrators in the sample of 100, including this female serial abuser, were known to be victims of domestic violence (as well as perpetrators).

Few distinguishing correlates

A key objective of the present study was to identify whether, once serial abusers were identified as such, their offending histories and personal backgrounds might reveal distinctive characteristics or profiles which could be used to inform not only protocols for their identification but also responses to effectively address their offending. Analysis of the bespoke database (n = 100) indicated that serial abusers do not appear to be a distinct subtype of domestic abuse perpetrators generally, or at least they cannot be identified as such using data derived from criminal justice agencies. For example, they are not significantly different from nonserial abusers according to their sex; race/ethnicity; age at time of assessment; age at first conviction; whether they have parenting responsibilities; and alcohol, drug, or mental health issues. As will be discussed, the emerging picture of this group is one of dissimilarity rather than similarity. Although further research on a larger sample of perpetrators is warranted before drawing firm conclusions, these preliminary findings challenge the efficacy of responses requiring their consistent and accurate identification.

More detailed information about those 13 perpetrators out of 100 judged by practitioners to be serial abusers is provided in Table 2. Their demographic characteristics were nearly identical to the population at large, as nearly all were male (12 out of 13) and white British (12 out of 13). Their average age was 35 years old. Apart from gender and age, however, the basic details contained in Table 2 indicate a fair degree of variation among this group of offenders. For example, as indicated by the number of probation assessments (OASys and SARA), some had fairly limited (known) offending histories—whereas others, such as “Harry” and “Sam,” were prolific offenders (both for domestic abuse as well as other types of crimes, both violent and acquisitive).⁹ Less than half (6 of the 13) seemed to be actively offending (i.e., were known to have committed domestic abuse within the past year). Two of the perpetrators (“Mike” and “Chris”) had quite dated offending histories, with no domestic abuse committed within the past three years (and in one case, much longer).

Analysis of their offending histories showed that most serial abusers were known to also have committed repeat offending (i.e., perpetrating domestic abuse on more than one occasion against each victim). Specifically, 10 of the 13 serial abusers (76.9%) were also repeat offenders, a higher proportion than nonserial abusers (58.6%), although this difference was not statistically significant (chi-sq. = 1.59; df = 1; p = .170).¹⁰ These 10 individuals, however, constitute only a small proportion of the total number of repeat offenders (10 of 61, or 16.4%). In other words, serial abusers appear more likely to be repeat offenders than vice versa, an area worthy of further study.

Although the data suggests that serial domestic abuse is likely to often involve repeat abuse against each victim, further analysis revealed that serial abuse cannot be equated with high-risk abuse. Contrary to expectations, only a minority of serial perpetrators (1 in 5) were considered to be at high risk of offending following the probation risk assessment process (i.e., using OASys and SARA). Specifically, 23.1% of serial perpetrators were considered at very high or high risk of harm according to OASys; 27.1% were deemed to be at high risk for committing violence to partners using the SARA tool; and 18.2% were believed to be at high risk for committing violence to others according to SARA. Furthermore, these percentages were not significantly different from nonserial abusers for OASys (chi-sq. = .86; df = 3; p = .836) or SARA risk to partners (chi-sq. = 2.1; df = 2; p = .350). However, a significantly greater proportion of serial abusers were rated as high risk to others according to SARA (chi-sq. = 6.0; df = 2; p = .050). These preliminary findings do not suggest a strong and direct relationship between serial and high-risk offending, although further research is warranted before drawing firm conclusions.

Finally, analysis of the individual SARA risk factors also contributes to the limited evidence base available for serial abusers specifically (see Table 3). A higher prevalence for 15 out of 20 items was found when comparing serial to nonserial abusers; however, only 4 of 20 comparisons were statistically significant (using the conventional standard of p < .05). Specifically, the assessments of serial abusers contained more evidence of past assault of family members (chi-sq. = 10.53; df = 2; p = .005), past assault of strangers and/or acquaintances (chi-sq. = 8.74; df = 2; p = .013), and past sexual jealousy and/or assault (chi-sq. = 7.39; df = 2; p = .025). In addition, serial abusers were significantly less likely to have recent employment problems (chi-sq. = 11.25; df = 2; p = .004). Interestingly, serial abusers did not seem to suffer disproportionately from mental health problems (e.g., suicidal ideation/intent, psychotic or manic symptoms, personality disorders, etc.). In combination, the information presented in Tables 2 and 3 reveals few distinguishing characteristics that could be reliably used to discriminate between serial and nonserial abusers. A larger sample size might have produced more significant differences between the groups, revealing a particular criminogenic profile of consistent predictors that could be used to aid in the effective identification and management of serial abusers, and this is an area worthy of further research. The implications of the current study for crafting new approaches for managing serial abusers, should such efforts be deemed helpful in the further development of coordinated community responses to domestic abuse, are discussed in the next section.

Table 3. SARA risk factors for serial versus nonserial domestic abusers.

Risk factor	Nonserial abuser	Serial abuser	P-value
Past assault of family members	14.5%	27.3%	.005
Past assault (strangers or acquaint.)	25.5%	27.3%	.013
Past breach	27.3%	27.3%	.768
Recent relationship problems	39.3%	54.5%	.643
Recent employment problems	25.5%	9.1%	.004
Family violence as a child	18.2%	36.4%	.257
Recent substance abuse	18.2%	45.5%	.112
Recent suicidal or homicidal intent	1.8%	9.1%	.436
Recent psychotic symptoms	1.8%	0.0%	.905
Personality disorder	9.1%	18.2%	.276
Past physical assault	47.3%	72.7%	.111
Past sexual assault	9.1%	9.1%	.025
Past use of weapons, death threats	5.5%	9.1%	.293
Recent escalation (freq. or severity)	14.5%	36.4%	.205
Past violation of no contact orders	9.1%	18.2%	.564
Denial of spousal assault history	14.5%	9.1%	.415
Attitudes supportive of DV	7.3%	9.1%	.941
Severe and/or sexual violence	5.5%	9.1%	.881
Use of weapons, death threats	7.3%	18.2%	.400
Violation of no contact order	5.5%	18.2%	.226

Note: Chi-square analyses based on n = 66–67 cases. Percentages represent the highest category for each risk factor (coded 0 = none, 1 = some, 2 = more).

Possibilities for management

The limited evidence from this exploratory research does not suggest that serial abusers represent a qualitatively different group—one that has a distinctive profile, can be reliably identified, and consequently can be recommended to receive a particular type of response. This is not to say that a “power few” highly dangerous serial abusers do not exist, but rather that current data collection and monitoring systems accessed for this research do not seem to be capable of reliably identifying them. Perhaps reflecting these difficulties, only two of the four Welsh police forces targeted serial abusers specifically within their force operational processes, thus highlighting that not all forces necessarily viewed serial perpetration as synonymous with the highest priority offending.

Many respondents desired a more offender-focused approach in order to effectively respond to domestic abuse and, ultimately, felt this was needed to break the cycle of domestic abuse. For example:

I’m sure we’re missing a trick with perpetrators … and we’re putting too much onus on our victims. We’re not exploiting as many mechanisms and opportunities with the perpetrators as we could. We can’t break the cycle for the victim unless we get something in place and deal with the perpetrator.

I can see a benefit in having a structured means of identifying domestic abuse perpetrators which may help agencies target them appropriately and ensure everything available is offered… . at present the support in place for perpetrators to address their behavior is only really postconviction and this is not ideal.

The effective management of serial abusers specifically, and how this fits into a more offender-focused response to domestic abuse, was less clear to respondents. How should information about serial abusers be meaningfully shared and utilized? For example, one respondent highlighted the challenge of linking information about serial domestic abuse to specific operational practices.

Once we decide we are happy labelling these people we then have to do something with them. A flag to say someone is serial, well what does that mean? Say the control room tells an officer on the way to a call that the individual has offended previously against five different women, what difference does that make for me knocking on their door now? We have that subtle knowledge that these people are out there and we go and do disclosures, but if we are going to overtly label someone as a serial perpetrator … What do you do with that information?

Concern was expressed by most respondents that individuals may be labelled as serial abusers without having a corresponding process of risk management or intervention to change their behavior in place. As several respondents acknowledged, unless a crime has been committed and the perpetrator convicted and subject to certain conditions and/or rehabilitative programs as part of their sentence, little else is available.

You lift the stone, and you label them, what are you going to do with them now? All we are doing is waiting for them to commit the offense again and then we deal with them … if we label someone, then what? Yes, I see the value, but all the other agencies need to own it too and provide the opportunity to refer these perpetrators [to support] in the same way we refer the victims.

There’s nothing to stop us saying to a perpetrator that you will comply with A, B, C and if you don’t there is the potential for X or Y to happen. This isn’t threatening, it’s building an action plan for a particular individual, and some will be compliant, some will want to comply. If we get them early enough … and this is where it’s not all about serial perpetrators, it’s all about intervention work … if you get them at age 21, 22 and first offense, building in some early intervention, we can stop them being our serial perpetrators of the future.

Overall, practitioners were able to articulate a range of challenges that would need to be remedied before any specific approaches for the management of serial abusers could be developed. Interestingly, notions of prevention, offender rehabilitation, and early intervention were often mentioned, even in the context of responding to the “power few” domestic abusers.

Conclusion

This exploratory study contributes new evidence about the extent and nature of serial domestic abuse. Like other research on serial criminality, it illustrates the complexity of adequately defining and responding to this problem (Edelstein, 2016; Kocsis & Irwin, 1998). The preliminary findings presented here question the prevailing assumption that serial domestic abusers, by sole virtue of their serial offending, are worthy of enhanced intervention and management. Thus, policies developed upon the unsubstantiated assumption that serial domestic abuse is inherently more harmful (then, for example, repeated escalating abuse against one victim or other high-risk offending that is not perpetrated against multiple intimate partners) must be greeted with caution. It will be necessary to first address the research gap on victim sequelae following serial versus other forms of domestic abuse before coming to a conclusion. Until then, this exploratory research highlights that inaccurate decisions and wasted resources are likely to follow from policy makers and practitioners making interchangeable references to serial, repeat, and high-risk domestic abuse as if they are synonymous. Instead, the current analysis indicates that there is only partial overlap among these three domains of offending. For example, only a minority of serial abusers were considered to be at high risk of further offending. Consequently, each domain (serial/repeat/high-risk) is useful for understanding a perpetrator’s pattern of behavior, and ultimately determining what should be done about it. In short, the overarching policy implication from this study is that efforts to target resources towards the “power few” domestic abusers should include information about serial alongside repeat and high-risk offending.

Although the term “serial abuser” inevitably invokes comparison with other notoriously dangerous types of serial offenders such as serial murderers and serial rapists, it is important to reflect on the implications of such connotations, which are inaccurate due to differences in scale. Thankfully, serial murder and rape are rare crimes; unfortunately, serial domestic abuse is not. As discussed previously, it has been estimated that 25,000 serial abusers are known to British police at any one time (Association of Chief Police Officers, 2009), and extrapolating from that report as well as the current study would suggest a figure of several thousand serial abusers in contact with Welsh criminal justice and partner agencies. Because estimates rely on official data, the true number is likely to be much higher. We must therefore carefully consider whether it is even possible to effectively monitor these individuals, assuming they could be reliably identified, and the substantial time commitment and practical implications for the involved agencies must be weighed against other possible uses of their finite resources.

Some lessons may be learned from areas that have implemented multiagency strategies focused on particular groups of high-volume offenders. Known as offender-focused deterrence strategies, they combine the use of a variety of criminal justice sanctions (“pulling levers”) with the provision of other services and resources to deter reoffending. Research has indicated the promising results of such strategies for reducing offending related to guns, gangs, and drugs in certain cities in the United States (see Braga & Weisburd, 2012; Kennedy, 1997), and they are now being applied to domestic abuse (see Kennedy, 2012, chapter 10). For example, an initiative in High Point, North Carolina, uses a focused deterrence approach to target and respond to the most serious domestic abusers (defined according to their criminal conviction record), in addition to offering a range of services to victims (Office of Community Oriented Policing Services [COPS], 2014). The first two years of implementation data show an overall reoffense rate of only 9% among more than 1,000 offenders as well as significant reductions in intimate partner homicide. Such approaches have recently made inroads in the United Kingdom, with initiatives currently being developed and tested in the Essex, Hertfordshire, Merseyside, South Wales, and Sussex police force areas (Houses of Parliament, 2015). One notable example is the Drive project, which aims to provide a combination of support and “disruption” tactics on a one-to-one basis with the perpetrators of high-risk victims (see http://www.safelives.org.uk/node/775).

In conclusion, initiatives that demonstrate effectiveness seem to be able to accurately identify and reduce the offending of the “power few”—the small percentage of offenders that cause a high percentage of harm (Sherman, 2007). One British study, relying upon police data, suggests that 80% of domestic abuse harm is attributable to less than 2% of victim-offender dyads (Bland & Ariel, 2015). As inferred by the current study, those that cause the most harm are likely to include some combination of serial, high-risk, and repeat domestic abusers. Further research on the profile of the “power few,” and the nature and context of their offending, should underpin the development of more intensive offender-focused responses to domestic abuse.

Notes

1. Some examples of serial domestic abuse reported in the popular British press include the following: http://www.bbc.co.uk/news/uk-scotland-glasgow-west-14489626, http://www.heraldscotland.com/news/crime-courts/four-years-for-serial-domestic-abuser-who-strangled-victims.25968701, http://www.dailymail.co.uk/news/article-2683609/Serial-domestic-abuser-murdered-fianc-e-single-punch-row-taxi-watched-fall-floor-like-bag-potatoes-jailed-life.html.

2. After much campaigning by Clare Wood’s father, this case led to the establishment of the Domestic Violence Disclosure Scheme. Commonly known as Clare’s Law, this new scheme allows police to disclose details of an abusive partner’s past so that people can make informed decisions about their relationships (https://www.gov.uk/government/news/clares-law-to-become-a-national-scheme). Clare’s Law was implemented across England and Wales in 2014 and in Scotland in 2015.

3. A comparatively smaller number was deemed appropriate given that this agency commissioned the research. Given the exploratory nature of the study, the limited resources available were focused on capturing the perspectives of those working in other agencies.

4. The Offender Assessment System (OASys) is a risk and needs assessment instrument, developed jointly by Her Majesty’s Prison Service (HMPS) and the National Probation Service (NPS). It provides the basis for integrated offender management throughout all custodial and community sentences.

5. A request was made to a data analyst within Wales Probation Trust to identify a random sample according to the perpetrator’s unique Police National Computer (PNC) number. This was a true random sample that was not stratified according to geographic region or any other variable.

6. All data was shared and managed in accordance with the U.K. Data Protection Act 1998. A named representative from the research team and each third-sector partner was required to sign a Joint Working and Confidentiality Agreement in order to participate in the research. All data was submitted to the author in anonymous format.

7. Unlike some U.S. states, in the United Kingdom there exists no specific criminal offense for domestic abuse. Thus the U.K. government establishes the official definition of domestic abuse, which was revised following public consultation in 2013 (see https://www.gov.uk/government/news/extended-definition-of-domestic-violence-takes-effect).

8. The names in quotations are fictitious and generated for illustration purposes only.

9. It is not the case that younger perpetrators in this sample have not yet had time to acquire more extensive offending histories, as their ages were not correlated with the number of assessments in their files (neither OASys nor SARA).

10. Only one serial perpetrator had no evidence of repeat offending (i.e., he was known to have committed only one offense against each of his two victims). The other two perpetrators were considered to have “maybe” committed repeat offending, because the information within the files was not conclusive.