The Concept and Measurement of Interpersonal Violence in Specialist Services Data: Inconsistencies, Outcomes and the Challenges of Synthesising Evidence

Abstract

Interpersonal violence comprises a variety of different types of violence that occur between individuals, including violence perpetrated by strangers and acquaintances, intimate partners and family members. Interpersonal violence is a leading cause of death, particularly among young adults. Inconsistencies in definitions and approaches to the measurement of interpersonal violence mean it is difficult to clearly understand its prevalence and the differences and similarities between its different subcategories and contexts. In the UK, specialist services provide support for victim-survivors and also perpetrators of violence. As well as delivering frontline services, specialist services collect data on interpersonal violence, both routinely and for the purpose of research and evaluation. This data has the potential to greatly improve understanding of violence in the UK; however, several issues make this challenging. This review describes and discusses some of the key challenges facing the two types of data collected by specialist services. Key inconsistencies regarding conceptualisation and measurement are identified, along with the implications of these for the synthesis of data, including implications for researchers, service providers, funders and commissioners. Recommendations are proposed to improve practice, the quality of data and, therefore, the understanding of interpersonal violence in the UK.

1. Introduction

Interpersonal violence is a global human rights and public health issue (Rosenberg et al. 2006). It is one of three forms of violence defined by the World Health Organisation, along with self-directed violence and collective violence, and is a broad category involving the intentional use of physical force or power against other persons by an individual or small group (Krug et al. 2002). Interpersonal violence can include physical, sexual and psychological violence and deprivation or neglect (Kilpatrick 2004). Specialist third-sector services provide support for many (but not all) forms of violence that fall within the broader remit of interpersonal violence, primarily domestic and sexual violence and abuse (DSVA). As such, specialist services are a source of information about certain types of interpersonal violence. The measurement of violence by specialist services is varied, with no real consensus on what the best methodologies are. Indeed, in a recent systematic scoping review of the outcomes utilised in evaluations and reports of support services and interventions for those who have experienced DSVA, 11 distinct categories of outcomes and 426 outcome measures were identified, of which fewer than half had been used in more than one publication (Carlisle et al. 2023). This illustrates the inconsistencies within the sector in terms of both what and how to measure when it comes to violence.

Interpersonal violence constitutes everyday violence such as sexual and physical assault and is one of the most common forms of violence against women (Montesanti 2015). Interpersonal violence includes family violence (child abuse, intimate partner violence and elder abuse), community violence (e.g., stranger rape and muggings) and institutional violence (e.g., abuse of inmates) (Mitchell and Anglin 2009). Exposure to interpersonal violence increases individuals’ lifelong vulnerability to a broad range of emotional, behavioural and physical health problems (Mercy et al. 2017). Improved measurement of interpersonal violence is important for strengthening the evidence base needed to increase the effectiveness of interventions and build a theory of change that can identify potential points of intervention. The challenges of measuring interpersonal violence include inconsistent definitions of violence and abuse across contexts and sectors, different sectors using different measurement frameworks and data not being consistently disaggregated by variables such as gender, ethnicity and disability, making it difficult to identify how interpersonal violence effects different groups. Were all data collected disaggregated by these and other important demographic characteristics, a more accurate intersectional picture of how interpersonal violence impacts population groups differently could be built.

While most current statistics on interpersonal violence derive from administrative authorities, an additional source of data that could be included, yet is currently underutilised, is that from specialist support services for victim-survivors and perpetrators. Specialist services commonly provide victim-survivor-centred support and safety, through specialist expertise in violence and abuse (European Institute for Gender Equality 2012). Possible services offered include refuges, outreach, counselling, legal advice, floating support, independent domestic violence advisor (IDVA) and independent sexual violence advisor (ISVA) support, children’s services and helplines (Council of Europe 2011b; Hagemann-White 2019; Floriani and Dudouet 2021; Macdonald 2021). Specialist support services tend to focus on DSVA and vary by ethos (traditional feminism vs. gender neutral (Taylor-Dunn and Erol 2021)); the specific population they serve (e.g., women only, men only and mixed gender and/or specific to Black and minoritised or Asian people); type of violence (e.g., domestic violence and abuse only, sexual violence and abuse only, other types of violence against women and girls (VAWG)) and whether they provide support for victim-survivors, perpetrators or whole families.

Specialist third-sector services produce two types of data that are relevant to the measurement of interpersonal violence: routinely collected administrative data and additional data collected for the purpose of undertaking research, such as evaluations. Administrative data are primarily collected by service providers for the purpose of supporting their day-to-day work and meeting the contractual requirements of funders and commissioners (Smith and Davidge 2022). Whilst not its intended purpose, the same study concluded that when administrative data are made accessible to researchers, they can “bolster their ability to respond expediently”, such that research findings can impact policy—in this case, responses to COVID-19 (Smith and Davidge 2022, p. 383). Administrative data are also often used to monitor and improve service provision (Kendall 2020) and are increasingly drawn upon for research and evaluation (Johnson and Stylianou 2022). Thus, service evaluations and impact reports may comprise purposively collected additional data, a combination of administrative data and purposively collected data or solely administrative datasets with no additional data from what is routinely collected.

This review aims to provide a brief overview of how specialist support services in the UK measure interpersonal violence, highlight some of the inconsistencies surrounding methods of measurement and discuss the implications of these inconsistencies in terms of the synthesis and linkage of data. First, the review considers how interpersonal violence is conceptualised by specialist support services and the impact this has on what is measured and collected. Next is a discussion of some of the main issues and inconsistencies with collecting additional data for research and evaluation, before moving on to consider the nature of routinely collected administrative data. The review concludes with a discussion of the implications for the measurement of violence in both types of data and some recommendations for researchers, service providers and funders.

2. Methods

We carried out a review of the literature in terms of the conceptual and theoretical studies discussing various issues surrounding interpersonal violence, its measurement and data within the specialist services context, as well as drawing on studies that have used this administrative and evaluative data; with a focus, where possible, on the UK context. This review stems from wider synthesis work being carried out in this area, primarily the aforementioned systematic scoping review that aimed to explore the outcomes currently being used by DSVA services and interventions when assessing effectiveness (Carlisle et al. 2023). Whilst carrying out this work, several challenges were encountered that form the basis for this review. A traditional literature review approach was deemed appropriate given the broad scope of the topic and the aim of providing an overview of the current state of the literature. Furthermore, had a systematic review been undertaken, many of the studies would not have made it into the paper because of not meeting the eligibility requirements; thus, opting for a narrative literature review enabled the inclusion of more studies from an area that is not published on enough.

3. The Conceptualisation of Violence in Specialist Support Third-Sector Services

Interpersonal violence encompasses a wide range of different violent and abusive behaviours (Krug et al. 2002). Specialist support services in the UK are often third-sector organisations with a focus on a particular subset or subsets of interpersonal violence in order to provide tailored specialist support. Thus, rather than all collecting data on all types of interpersonal violence, different specialist services will collect data relating to the type/s of violence they provide support for.

One such subset of interpersonal violence that is commonly provided for by specialist support services is domestic violence and abuse (DVA). In the UK, there are several definitions of domestic abuse. The 2021 Domestic Abuse Act defines it as “the act of any of the following: physical or sexual abuse; violent or threatening behaviour; controlling or coercive behaviour; economic abuse; psychological, emotional or other abuse. For the definition to apply, both parties must be aged 16 or over and ‘personally connected’” (UK Government 2021). The conceptualisations of DVA used by specialist services are largely congruent with this definition; they all consider all such acts to be violence, even those that are not ‘severe’ or not ‘physical’. However, there are slight differences in approaches that have implications for the data that are collected. For instance, whilst not initially included in the Domestic Abuse Act 2021 definition (but later added (The Crown Prosecution Service 2022)), many specialist services were already providing for and collecting data on violence against children, including those who have witnessed DVA. This is in line with recent proposed international classification of violence against children (United Nations Children’s Fund 2023). Gender is also not specified in the above definition; however, many DVA specialist services only cater for women, in response to the fact that most DVA is perpetrated against women (Walby and Towers 2017; Office for National Statistics (ONS) 2020), whilst a handful serve only male victim-survivors.

Other subsets of interpersonal violence that are the focus of different specialist services in the UK include sexual violence and abuse (SVA), so-called ‘honour-based’ violence, including forced marriage and female genital mutilation, and stalking. Some services provide support for victim-survivors of any form of VAWG, and some provide gender-inclusive DSVA services. Some are single focused, providing support for DVA or SVA only, whilst others are dual focussed, providing services for both DVA and SVA.

Such differences have implications for comparing, integrating and linking specialist services data. First, because different services provide support for different types of abuse, there is a possibility of double counting. For instance, whether or not SVA is a type of DVA depends on the relationship between the perpetrator and victim-survivor. People may have also experienced both DVA and stranger-perpetrated SVA. Therefore, service users who have experienced SVA and access a service catered to DVA may also get referred to and receive support from a specialist SVA service. Because specialist services operating within the boundaries of different local authority contracts often record different data in incompatible systems, the same individuals can appear in multiple datasets (Bowstead 2019). To accurately establish the effectiveness of interventions as part of a multisectoral response, the use of unique identifiers may be required to understand the impact of multiple visits to different services (Kendall 2020). Second, the parameters of the data must be considered and understood before any conclusions can be made. Different services have different criteria in terms of the population they support and the specific type of violence they collect data on. Thus, even if it were possible to collate data from all specialist services across the country, it might not be a nationally representative sample and, therefore, could not be used to calculate a precise prevalence of (certain types of) interpersonal violence. However, it would still provide useful insights into the context underlying different forms of violence against different subgroups of victim-survivors.

4. Methods Used to Measure Violence

The monitoring and evaluation of specialist support services can be a contentious issue, with concerns regarding compromising victim-survivor safety, requiring inappropriate outcomes, and worry that findings may be used against services (Sullivan 2011). On the other hand, it enables such services to evidence their effectiveness, which is key to securing and maintaining funding. It also facilitates research so that the most effective methods of providing support can be identified, used to develop service provision and address gaps and unmet needs for the population(s) they work with, and provide a mechanism through which specialist services can contribute to the knowledge and evidence base on VAWG (Imkaan et al. 2016). The methods on offer to researchers, evaluators and service providers to measure violence and the impact of services are wide ranging. Each type of methodology has strengths and limitations and is implemented to varying degrees across the sector. Some distinctions between the different methods that have primarily been used by specialist services are discussed below.

4.1. Time Points

The point at which an outcome is measured has implications for what can be inferred from the data. For instance, if the aim is to assess the impact of a service or services’ activity on violence, multiple time points are necessary, with one at baseline (i.e., before any support or activity takes place) and at least one at the end of service use. This allows inferences to be made regarding the impact of the service on the outcome. However, this is often challenging because of victim-survivors suddenly disengaging with services (i.e., dropping out) or only engaging on a one-off or brief basis (Campbell et al. 2008). An example of specialist service data being used to measure the impact of service(s) on violence in this way is the SafeLives Insights system, which collects data from intake to case closure (SafeLives n.d.). Continuing to measure outcomes beyond service use potentially allows for the assessment of long-term impacts to see if changes are sustained and for how long. However, these time points are particularly difficult for specialist services to measure because of issues around resources and capacity, safety and safeguarding (Sullivan 2011). Some services report data at a single time point. Often, this is framed retrospectively, asking service users to reflect back to when they first entered the service and report whether the outcome has improved or not (e.g., Advance 2021; Imkaan 2012). Whilst this is less resource intensive, it produces less reliable data, as this type of retrospective measure is prone to recollection errors and biases (Raphael 1987). For instance, in the context of psychotherapy, McFarland and Beuhler found that the majority of participants overestimated their pre-therapy levels of distress, resulting in an overestimation of positive change (McFarland and Buehler 1998). The accuracy of recollection can be influenced by many factors, including mood, individual characteristics and personality and cognitive biases (Stone and Shiffman 1994; McFarland and Buehler 1998; Safer and Keuler 2002). Therefore, whilst capturing data at multiple time points would be more burdensome for services, doing so can result in more accurate measures of the impact of the service on victim-survivor outcomes. Single time-point measures are adequate for providing a snapshot, for instance, when wanting to understand the demographics and violence histories of service-users when presenting to a service for the first time.

4.2. Qualitative and Quantitative Data

Specialist services collect both quantitative and qualitative data. Generally speaking, much of the administrative data collected by specialist services are quantitative, as data collection systems use structured forms and include descriptive demographic data such as age, gender, sexual orientation and ethnicity, as well as data surrounding violence, such as the relationship to the perpetrator, and the types and severity of abuse experienced, measured as categorical outcomes (e.g., SafeLives 2021). Many victim-survivors experience multiple types of abuse concomitantly, and all types of abuse disclosed are recorded (Women’s Aid 2021). However, such information is sometimes recorded in data systems as string variables in free-text fields, meaning considerable data management is needed to undertake statistical analysis (Green et al. 2015). Quantitative data collected by specialist services may also include continuous measures such as mean scores on the Beck depression inventory (Beck et al. 1987) or Rosenberg self-esteem scale (Rosenberg 1965), or categorical measures such as the severity-of-abuse grid (Campbell and Soeken 1999); which was the most commonly identified outcome measure in a recent scoping review (Carlisle et al. 2023). Such measures are more commonly administered for evaluations rather than being included in routine administrative datasets. Qualitative data are also commonly collected for evaluations. These data can be captured through interviews, focus groups, case studies and/or open-ended questionnaires. They capture rich data on service users’ experience of services, the barriers and challenges they face accessing them and the benefits they have gained as a result of them (e.g., Survivors Network and Switchboard 2021; Monkton Smith 2010; Solace Women’s Aid n.d.). Each type of data has its own advantages and limitations in this context. Quantitative data are often quicker and simpler to collect, combine and compare and are, therefore, more straightforward to interpret and draw conclusions from. However, they lack the richness of qualitative data, potentially masking the complexities and nuances that often underlie interpersonal violence. Qualitative data provide detailed narratives that allow for a deeper understanding of individual experiences, meaning they have much to offer the policy community (Natow 2022), although they are more time consuming to collect and analyse. Thus, whilst both types of data are valuable for service providers and for understanding experiences of violence, qualitative data tend to be less valued by policymakers, who prioritise statistics for demonstrating measurable changes in violence (Office for Statistics Regulation n.d.; Connelly et al. 2016).

4.3. Outcomes

Violence can be measured in numerous ways. The specific violence outcome(s) measured by services are informed by the priorities and goals of the service and/or funders and by the type of violence the service is targeted at. For instance, violence could be measured as the average number of violent incidents experienced by victim-survivors; however, in cases of coercive control where distinct incidents may be difficult to define and patterns of abuse over time are more relevant, this may not be an appropriate measure. Instead, it may be more appropriate to assess outcomes relating to the pattern of abuse, such as the severity of abuse, its frequency, the number of perpetrators and the relationship with them, to better understand the context within which the violence occurred. Thus, whilst all are under the umbrella of violence, the specific outcomes measured by services differ and may not be directly comparable. Essentially, the responsibility for the cessation of abuse lies with the perpetrator and not with the specialist support services or the victim-survivors they support (Sullivan and Bybee 1999; Stark 2007). Therefore, whilst some funders require the measurement of outcomes such as leaving the abusive relationship or the cessation of abuse, many argue that these are not the most appropriate outcomes, as they are not in the control of the victim-survivor and may actually cause an escalation in abuse (Sullivan 2011; Stark 2007). Thus, as well as measuring violence directly, relevant outcomes may also include indirect measures of violence. Indeed, interpersonal violence can impact numerous aspects of a person’s life, including physical health, mental health, social relationships, employment and finances. Therefore, outcomes relating to autonomy and empowerment, mental wellbeing, support networks and establishing independence may be more appropriate ways to assess and evaluate specialist services whilst providing indirect measurement of violence and its impact.

4.4. Outcome Measurement Tools

Once relevant outcomes to measure have been identified, there are a range of different ways measurement can be operationalised. For example, a review of validated measures found 19 different measures of mental wellbeing and mental ill health (Taggart and Stewart-Brown n.d.). Therefore, services wanting to assess mental health have a wide range of available options, leading to further discrepancies between services. Furthermore, specialist services often do not use standardised measures, opting instead for single-item measures that are also often inconsistent across services. Indeed, the previously mentioned scoping review identified 283 different outcome measures used in evaluations and reports of support services and interventions, which included validated questionnaires, unvalidated questionnaires and single-item questions (Carlisle et al. 2023). Whilst complete uniformity is not plausible because of differing prioritisation across the sector, improved agreement on the specific measures used would be beneficial in aiding comparisons and syntheses.

The benefits of using standardised outcome measures are that they have been shown to reliably measure what they set out to measure and are often accompanied by clinical cut-offs indicating when an individual may be in need of professional help. For instance, the Beck depression inventory (Beck et al. 1987) is a 21-item questionnaire with scores ranging from 0–63, which provides cut-offs indicating the severity of depression (i.e., severe depression is indicated by a score of 30 or over). However, such questionnaires can be time consuming, both to administer and to score. Thus, whilst such outcome measures are common in research, they are less commonly used by specialist services. Single-item measures (i.e., a single question: “are you experiencing depression”) are more frequently seen in evaluations of specialist services. Such measures are less time consuming but give blunter data on the outcome in question. Single-item questions may also make comparisons and syntheses across different services or publications more challenging, as slightly different phrasing or possible outcome responses result in differences in what is actually being measured. A particular challenge in measuring violence is that there is a lack of validated measurement tools specifically assessing violence. A few exist (e.g., the conflict tactics scale (CTS); Straus 1979), but they are seemingly rarely utilised. For instance, of the 426 outcomes and 87 interventions and services identified in the previously described scoping review, none utilised the CTS (Carlisle et al. 2023).

4.5. Self-Report or Objective Outcomes

Outcomes can also differ in terms of their subjectivity. Examples of purely subjective measures include victim-survivor perceptions of safety and wellbeing. Subjective measures have the benefit of being able to reflect how victim-survivors have experienced violence; however, some caution should be exercised, as they are reliant on memory, which may not be accurate, are vulnerable to socially desirable responding (by perpetrators) and are open to interpretation. Research on violence often utilises retrospective self-report questionnaires, which have produced a wealth of knowledge about interpersonal violence (Grych and Hamby 2014). However, reliance on a single type of measurement has constrained what is studied in violence research, limited understanding of important aspects of interpersonal violence and produced longstanding controversies such as questions about gender patterns in intimate partner violence (Hamby 2014). Objective measures of violence, such as those that can be derived from police data (e.g., the number of police callouts or 999 calls), are potentially more reliable and accurate as they do not rely on interpretation, but they do not capture the whole picture because not all incidents of abuse will be reported to police (Office for National Statistics 2021).

5. How DSVA Services Measure and Record Violence in Administrative Datasets

At the level of services and interventions, there are both similarities and differences in the way that providers measure the violence experienced by victim-survivors accessing their services, and the impact of the various forms of support provided. Whilst there is significant overlap in administrative data collected between services, there is also variation and modification to standardised measurement tools.

5.1. What Is Measured within Administrative Datasets and How It Is Recorded

Most specialist support services collect data from victim-survivors at intake, reviewing information periodically throughout the support period and again when the victim-survivor exits the service. Commonly, they record information about any violence experienced (not just the most recent or severe incident and including experiences of multiple abuse types at the same time), its impacts, support received and service- and individual-level outcomes. Such data are usually recorded in case management systems and/or computerised case notes, although the sophistication of data recording and reporting systems varies across services, largely as a function of size and resource capacity (Smith and Davidge 2022). Datasets typically include sociodemographic information, history of abuse, information about perpetrator/s, client needs, support provided, various service- and individual-level outcomes, agencies involved, criminal and civil justice outcomes, client feedback and experiences of external services (criminal justice, civil justice, health and other DSVA services) (Walby et al. 2017). They also tend to include individual demographic variables less explored in research, such as sexual orientation, gender identity, religion and disability. Much of the data collected are recorded as categorical structured variables; however, narrative accounts of the experience of violence and associated circumstances are collected in free-text form (Walby et al. 2017). Data are primarily recorded by frontline staff, which can enhance data collection because of the trust and rapport they build with victim-survivors (Smith and Davidge 2022). Most of the information collected is from the victim-survivor’s perspective, although as understood by the service providers recording such information.

Whilst statutory agencies such as the police and health services tend to count specific incidents of violence (or events), the unit of measurement in specialist services data has traditionally centred around the victim-survivor (or perpetrator, in the case of perpetrator programmes). This is because support is based on their needs; thus, recording individual events may not be the priority or appropriate. Measures of intimate partner violence that are time bound and incident specific can mask the chronicity and severity of post-separation abuse (Spearman et al. 2023). As case management systems used by specialist services have evolved, many now use the victim-survivors’ journey through their services (i.e., the period of support) as the unit of measurement, meaning they are able to collate data on the survivors they support under unique identifiers. This practice allows services to input information about the victim-survivors’ experiences of abuse each time they start accessing a service, thereby documenting their history of violence and capturing repeat victimisation, experiences of multiple types of violence and violence involving multiple perpetrators. Recording and measuring DSVA in this way provides some opportunities for longitudinal analysis of experiences of violence and analysis of relationships between service use/refusal of referral/not using services, types of services used, charges and convictions, etc. and outcomes over time (Kendall 2020). Such records help to estimate the current level of known cases and identify trends, although the latter is more difficult to assess because changing policy priorities and recording practices and/or the increase or decrease in resources can change the profile of detected cases (Kelly and Karsna 2017). Furthermore, because data collection is often boundaried to the local authority areas or regions, rather than being nationwide, it is rarely possible to identify individuals moving between areas and services (Bowstead 2019). Collecting data on violence longitudinally potentially enables measurement of the cost of different types of violence using a lifetime approach. Such analysis can produce costs higher than previously published estimates, which focus on a single event, as demonstrated by Capelas Barbosa’s (n.d.) study of the costs of adult and child sexual abuse using Rape Crisis data.

Unlike many surveys of violence and abuse, which tend to be biased towards collecting data about physical forms of violence and abuse (Skafida et al. 2023), specialist support services collect data covering a broad range of abuse types. This means specialist services datasets capture types of violence often not picked up by surveys, which may not be immediately recognised as violence (although they fit into the important definitions (World Health Assembly 1996; Council of Europe 2011a; United Nations General Assembly 1994)) but do cause harm to another individual, such as economic violence, coercive controlling behaviours, honour-related violence, modern slavery and technology-facilitated abuse (Skafida et al. 2023). The ways that different types of violence are categorised are also not always consistent across services, which can make comparisons difficult. Nevertheless, by adopting a broader conceptualisation of the types of violence people may experience, specialist services are more likely to capture how gender and other inequalities and contexts of coercive control influence violent behaviours and their impacts. Improved system-wide understanding of this would enable us to better measure the full extent of the problem in different contexts (Mannell et al. 2021), which would in turn lead to more effective preventive action.

Specialist support services place as much focus upon measuring the impact of violence and abuse on victim-survivors as measuring the violence itself. This is because, whilst statistics can indicate how much of a problem DSVA is, they cannot show how it affects victim-survivors (Harwin 2006, p. 559). To meet their multiple needs for protection, support, empowerment and long-term safety, specialist services measure beyond physical injuries to the impact of DSVA on the lives and behaviours of victim-survivors, often including those who have witnessed it, who are considered victim-survivors in their own right. This has implications for the way the severity and frequency of violence is measured, by acknowledging that even non-physical, ‘low-risk/level’ behaviours can have substantial and wide-ranging impacts on victim-survivors and others affected, and that reductions in the recorded frequency of violent incidents does not necessarily mean reduced risk or severity of violence (Day et al. 2009; Myhill 2017). When considered in the context of sustained, systemic and insidious patterns of abuse, identifying and measuring whether and how often victim-survivors have experienced different forms of violence are not always feasible or appropriate (Sharp-Jeffs et al. 2018). In this sense, the severity of the violence is measured according to the breadth of harms caused, as opposed to the specific violent act(s) used to inflict those harms.

Due to a patchwork provision of specialist support services geographically in the UK, the piecemeal and insecure funding to support the collection and analysis of data and the associated issue of inconsistent data across specialist services, synthesising specialist services administrative data on DSVA to be analysed on a national scale is difficult (Kendall 2020).

5.2. Who Is Included

Specialist support services capture violence experienced by population groups who are routinely left out of sampling frames for most national surveys and those who are less likely to report the violence to police or health agencies because of distrust and other reasons, such as those living in sheltered accommodation, those changing addresses very frequently, those living temporarily with friends or family, homeless people and those with insecure migration status (Voolma 2018; Skafida et al. 2023). Due to sampling and difficulties of access, surveys normally do not include those populations at highest risk of experiencing interpersonal violence. This has clear implications for the accuracy of survey measurements of the prevalence of interpersonal violence.

Specialist services tend to collect more data on perpetrators than other sectors and agencies, beyond just the victim-perpetrator relationship (Kelly and Karsna 2017). This can include the relationship of the primary and any additional perpetrators to the victim-survivor, whether the victim-survivor lives with the perpetrator, child-contact arrangements, the perpetrator’s sex and/or gender and information about perpetrator needs in areas such as mental health, alcohol and drug use and charges or convictions brought against the perpetrator (e.g., SafeLives 2019). Because experiences of violence are typically measured before, during and after support is provided, changes in victim–perpetrator relationships and the nature of the contact between victim-survivors and perpetrators can be recorded, enabling the measurement of post-separation abuse (e.g., Hester et al. 2020). Collection of richer data on perpetrators and perpetration of DSVA (by services that provide perpetrator programmes) means that the sources of support accessed can be analytically associated with a perpetrator or type of abuse experienced, links that cannot be made in other sources of data on violence. Some specialist services support both the victim-survivor and perpetrator at the same time, or whole families, meaning changes in both perpetration and victimisation can potentially be measured.

5.3. Measurement of Risk

Specialist services measure the risk of further violence to victim-survivors engaged with their services. In England and Wales, professionals are expected to assess risk level using a standardised tool, the domestic abuse, stalking and honour (DASH) risk assessment. DASH has been widely implemented in police and healthcare settings and is used to inform decisions made at multi-agency risk assessment conferences (MARACs), although how effectively it measures risk has been questioned (Turner et al. 2019; Friskney et al. 2021). However, even measures of risk across specialist services vary; whilst some use the standard DASH form, others have adapted it, and some have developed bespoke risk assessment tools. For example, by-and-for specialist services have developed complex and dynamic assessment tools to better capture the risk of family and community collusion in types of violence referred to as ‘harmful practices’ (such as forced marriage and honour-based abuse) (Jeffery 2023). Such assessments are reviewed on a regular basis to ensure measurement of emerging and escalating or changing risks, something the inflexibility of the CAADA-DASH limits (Munro and Aitken 2020). Research emphasises the need for culturally sensitive risk assessment and service provision, particularly when supporting historically marginalised populations, in order to capture the complex dynamics that comprise social identities, social contexts, vulnerabilities, strengths and histories, all of which shape both the individuals’ experience of victimisation, and others’ social responses to them (White et al. 2019). Such an intersectional approach underpinning risk assessment practices would enable more accurate measurement of the violence experienced by those from underserved, marginalised and culturally specific populations.

6. Discussion: Implications for DSVA Services’ Data Collection and Its Contribution to the Measurement of Violence

6.1. Implications for Evaluation

The issues highlighted here illustrate the inconsistent and variable nature of violence measurement by specialist support services as part of service monitoring and evaluation. There are variations in the approach to measurement, the specific outcomes that are measured, the tools that are used to measure them and when data are collected, all with implications for synthesis and comparison. For instance, variability in the specific outcome measurement tools, where there may be differences in scales used for continuous outcomes or in how similar outcomes are defined, means that grouping these outcomes together for meta-analysis is not appropriate. Different approaches to evaluation, including different time points, whether outcomes are collected prospectively or retrospectively, and whether evaluations include control groups, has implications for the specific tools that can be used for assessing study quality. For instance, to quality assess the 80 studies reported in Carlisle et al. (2023), at least four different tools would be needed. Thus, variability limits possibilities for synthesis and makes any synthesis that is feasible much more challenging.

One of the factors that is likely contributing to this inconsistency is a lack of consensus surrounding what should be prioritised when it comes to determining success (Westmarland and Kelly 2013). In addition to a lack of consistency and consensus across the sector, there is sometimes a lack of robust and rigorous methodology, likely a symptom of the limited time and resources services have to collect data, on top of their primary priority of supporting victim-survivors. One way to address the inconsistent nature of violence measurement in service evaluations is an agreement on a core set of outcomes, as has been developed in the area of child maltreatment (Powell et al. 2022). Core outcome sets should be co-produced, meaningfully involving people who have experienced interpersonal violence and who are accessing these services, as well as service providers, funders, commissioners and researchers, to ensure that the outcomes that matter the most to the people who matter the most are measured. A more inclusive and shared conceptualisation of how to measure success would reduce inconsistencies and lead to a stronger evidence base, allowing more easy synthesis and linkage.

Additionally, there needs to be a greater focus on the development and validation of measurement tools. Validated tools are needed to assess various types of violence (e.g., sexual assault) and to evaluate outcomes (e.g., instruments to assess the efficacy of risk assessments). This would improve consistency in not only what outcomes should be measured but also how they are measured.

Finally, efforts to improve consistency in how specialist services measure interpersonal violence must be balanced with efforts to improve the measurement of violence more broadly as our understanding of the phenomenon evolves, and openness to novel or emerging approaches to intervention. Consistent outcome measurement is not helpful if the outcomes are no longer relevant or appropriate.

6.2. Implications for Service Providers

Research suggests a consensus between frontline service providers, survivors, commissioners, policymakers and researchers on the value of collecting data about experiences of violence in a consistent and reportable way (Christie and Karsna 2019; Powell et al. 2022). The data collected can aid the development of service provision and efficient resource allocation and feed into the developing knowledge base on violence. However, whilst there may be benefits of a shared multisector core measurement framework that can be used by different agencies, it is vital to remain sensitive to the unique importance of different specialist services (Imkaan et al. 2016). In such a resource-constrained setting, there is also a need to strike the right balance between the costs of collecting information that may not be analysed regularly or useful on a daily basis versus dedicating limited resources to investing in a small, but good-quality, minimum dataset and in direct service provision (Kendall 2020). Service providers will and should always put victim-survivors’ safety, privacy and confidentiality first, meaning efforts to create national data-management systems and registries that include data on violence against women and link them can be a cause for concern (ASEAN 2018).

6.3. Implications for Funding and Commissioning

When used effectively, specialist support services’ administrative data can provide rich findings to contribute to the evidence base on DSVA and inform practice and policies in response to these forms of interpersonal violence. Administrative data can create opportunities for detailed analysis (Hurren et al. 2017), which in the case of specialist DSVA support services could provide nuanced insight into the nature and scope of violence that is not currently captured in national statistics. However, the value of specialist services data is not fully exploited, largely because lack of funding and resources limits the extent to which such data can be interrogated in creative ways. Crucially, specialist by-and-for services that serve minoritised people are disproportionately underfunded (Imkaan 2018), meaning they are less likely to have data infrastructure in place, with implications for the measurement of experiences of violence within certain high-risk populations. Variation in the quality of data systems means some services’ data lend themselves to external evaluation and to providing evidence for policy making more easily than others. Thus, the extent to which specialist services data can meaningfully enhance the measurement of interpersonal violence is limited by the chronic underfunding of the sector. Enhancing data collection and recording practices to better measure the violence experienced by those who seek support requires additional committed funding and resources, as their priority is providing frontline services that best meet the needs of victim-survivors.

7. Recommendations

• There is an overall need for more consistency in how interpersonal violence is measured within specialist services. Given their substantial influence over what data are collected, consistency across commissioning frameworks could help with this.
• A core outcomes framework needs to be co-developed with a range of stakeholders, including people with lived experience. This should identify the key outcomes that mark ‘success’ and ideally have agreement on specific measures and/or tools and recommended methods for measurement.
• Methods for linking specialist services data with other sources of administrative data on violence (such as the health and criminal justice fields) whilst protecting the privacy of individuals should be further explored. Collaboration between specialist services and academic researchers can ease the burden of conducting such complex analyses, but it is paramount that any such research is co-produced, so that the perspectives of service providers and those accessing services are embedded.
• There is a need for sustainable funding within the third sector, including for smaller services. The current piecemeal and precarious nature of specialist services funding does not lend itself well to consistent data collection and evaluation. Secure funding that allows providers the time and resources for data collection will enable better-quality data and more rigorous evaluation.