top of page

Who commits domestic abuse? The lessons of our research for preventing serious violence

Consulting Report


Ellie Covell, Head of Operations

Wednesday 22nd February 2023


Last year Crest Advisory was funded by the Home Office to conduct research into the offending patterns of domestic abusers in the West Midlands. We are now publishing this research, along with the key implications for practitioners working with domestic abuse perpetrators and those at risk of offending. Our findings are timely as public bodies in each area across England and Wales are required to work together to prevent serious violence as part of a new statutory duty.

The Serious Violence Duty came into effect at the end of January. Each local area can develop its own definition of ‘serious violence’, as long as it’s based on evidence. Crest’s work, helping local agencies prepare for and implement the duty, suggests domestic abuse should be considered in local area strategic needs assessments, as it is sometimes an indicator of further serious violence.


In the second round of the Government’s domestic abuse perpetrator intervention fund, the Home Office identified seven priority areas for research. Crest and West Midlands Police (WMP) put forward a joint proposal to use police data to better understand trajectories of offending among those committing domestic abuse offences. We had worked with WMP before on an analysis of domestic abuse demand on police and believed our proposed research could help prevent further incidents.

The study was primarily quantitative. We analysed 660,000 crime records over a ten-year period, involving 178,000 victims and 145,000 suspected perpetrators of domestic abuse. We conducted a ‘clustering’ analysis which grouped together offenders with similar characteristics and we carried out ‘regression’ analyses to understand the relationship between offending patterns, risk levels and the outcome of crimes.

We also questioned people working in the criminal justice system and the voluntary sector, while in-depth interviews were carried out with two individuals who had committed a number of offences, including those relatd to domestic abuse. That helped put the data into context.


We wanted to explore three questions:

  1. Would there be any benefit in practitioners identifying domestic abuse risks through the non-domestic offending trajectories of offenders?

  2. To what extent does wider offending seem to be associated with domestic abuse offending?

  3. What could wider offending histories/behaviours/experiences tell us about different types of domestic abuse perpetrators?


Most of the harm from domestic abuse was caused by a relatively small group of individuals

We found that over the period we were looking at, April 2011 to the end of March 2021, the majority of individuals listed as a domestic abuse suspect, had just one lower harm crime or incident recorded against them (which generally the police had graded at the lowest risk level). A relatively small group of individuals had multiple offences listed against each of them; they were responsible for most of the harm from domestic abuse.

High-harm individuals were more likely to have a history of other offending unrelated to domestic abuse

People who caused the most domestic abuse-related harm committed more non-domestic abuse offences than others. There were some exceptions - in particular a group of high-harm offenders who only had a history of domestic abuse and a relatively small number of offences recorded against them. But these offences tended to indicate a pattern of sustained behaviour, for example, coercive control offences which had gone unreported for several years.

The non-domestic offending associated with domestic abuse offenders was often abuse of children or vulnerable adults (in line with findings from previous research). However, our research also highlighted other offences that could be understood as a risk factor for domestic abuse.

There was a link between certain non-domestic abuse crimes and higher-harm domestic abuse - 16 per cent of the non-domestic abuse-related offence types in the analysis had a statistically significant positive effect on the domestic abuse level of harm [1]. For example, the level of harm committed by a domestic abuse offender who had also carried out robberies was likely to be higher than the harm caused by someone without that kind of offending history. We found evidence of a relationship between domestic abuse and non-domestic abuse across four broad categories:

  • Sexual offences: rape and sexual offences against both adults and children

  • Acquisitive crime: attempted burglary, robbery, making off without payment and theft of motor vehicles

  • Violence: malicious wounding, grievous bodily harm, racially aggravated assaults, threats to kill and threats with a weapon

  • Anti-authority: breaches of court orders, such as non-molestation orders, bail and licence conditions; assaults on police; threats to witnesses and jurors; criminal damage and driving offences

In line with previous research, perpetrators are not one distinct group

Previous academic studies [2] have divided domestic abusers into groups according to the severity of their offending and the different types of offences they committed.

We have carried out a clustering analysis [3] in a similar way, shown in the diagram below. The size of the bubbles represents the amount of harm caused by offenders. Those in group B cause the highest harm: they carry out more offences - crimes that are related to domestic abuse and offences which aren’t.

Domestic abuse offenders grouped according to the volume and severity of offending

Within each group, we found some relatively distinct sub-groups with different characteristics. Group B is of particular interest because this is where we saw offenders with both the highest rates of domestic abuse offending and non-domestic abuse offending.

Within Group B, we found six sub-groups (outlined fully in the research report):

  • High-harm offenders: A very small group of offenders with extremely prolific and high-harm domestic abuse profiles.

  • Economic compulsive offenders: A lower-harm group tended to commit a lot of theft-related crimes and a smaller number of domestic abuse offences. A second group caused significant domestic abuse harm alongside lower volumes of acquisitive offences.

  • Prolific violent offenders: On average, they were involved in 20 domestic abuse incidents and 12 non-domestic abuse incidents - which tended to be violent.

  • Prolific young offenders: The average age of offenders in this group was much younger, with an offending profile more heavily weighted to non-domestic abuse offences.

  • Non-specialised law breakers: Mostly male offenders who, on average, committed 15 offences against eight different victims - with no clear offending type or pattern.

Some of the categories and descriptions are already well understood and resonated with the practitioners we interviewed. The ‘prolific violent offender’ group for example was often mentioned:

"I can definitely think of cases where they’ve got offences against members of the public, and then in domestic settings they perpetrate violence, or physical violence to a partner. [...] In either scenario, they don't really differentiate between who the victim is - it's more about releasing their aggression”

- Multi-agency risk assessment conference (MARAC) stakeholder.

The offending history of one of our research participants, David [4], aligned with that of the ‘prolific violent offender’. David committed violent offences in both domestic and non-domestic settings, suggesting that he had normalised violence as a means of dealing with conflict. He received two custodial sentences for non-domestic abuse-related violent offences while under the age of 18. Later in life, David’s domestic abuse offending was also very violent.

Summary of David’s background and history of offending

David helped us to understand how our research could help practitioners identify possible intervention points and align some of the risk factors for offending across both domestic abuse and wider offending. In David’s case, an earlier mental health diagnosis may have prevented some of both the domestic abuse and non-domestic abuse offending.

Implications and impact

The criminal justice system is clearly unable to cope with the volume of domestic abuse victims and offenders, so cases must be prioritised. Grouping perpetrators according to the number and nature of the offences they’ve committed helps identify those at the greatest risk of causing harm. That means resources can be targeted effectively, which is vital for agencies at each stage of the justice system.

Our research found that close to two-thirds of all domestic abuse perpetrators in the West Midlands had little, if any, history of offending that practitioners could have worked on. However, those who did have an offending history presented a higher risk of harm when they commits domestic abuse offences. That has led us to make a series of recommendations in our report, to improve risk assessments and help agencies intervene to prevent domestic abuse.

  • Recommendation 1: Wider offending history should inform domestic abuse intervention

  • Recommendation 2: Identify specific non-domestic abuse offences that factor into harm

  • Recommendation 3: Breach offences should trigger enhanced assessment and response

  • Recommendation 4: Offences indicative of wider abuse tendencies should be flagged

  • Recommendation 5: Substance misuse and domestic abuse should be tackled in tandem

  • Recommendation 6: Risk assessment processes should take account of previous risks

There is an additional, top-level, recommendation which we hope will be addressed as a matter of urgency to prevent domestic abuse and other crimes.

Offending history should be an important factor in the design and operation of criminal justice processes and when decisions are taken to commission and provide services.

West Midlands Police, as key contributors to this research, had early sight of the findings and have already started to respond to the recommendations around the initial police response. In particular, a bid for funding to support a programme to assess and refer domestic abuse perpetrators while they are in police custody has been submitted to the Home Office for consideration.

WMP Head of Domestic Abuse Tony Hopkins said: “The research has been invaluable to reviewing our internal processes and helping inform a bid to the Home Office to specifically tackle domestic abuse perpetrators when they are in custody.”



[1] When using the standard p<0.05 significance level

[2] Cunningham, A., Jaffe, P.G., Baker, L., Dick, T., Malla, S., Mazaheri, N. and Poisson, S., 1998. Theory-derived explanations of male violence against female partners: Literature update and related implications for treatment and evaluation (pp. 1-10). London: London Family Court Clinic.

[3] The clustering analysis was based on 37 factors including: age of perpetrator and how it compared to age of victim(s), incident type, severity and frequency of offending, and location of the incidentThe clustering analysis was based on 37 factors including: age of perpetrator and how it compared to age of victim(s), incident type, severity and frequency of offending, and location of the incident

[4] Not his real name


bottom of page