Research and analysis

Violence Reduction Units, year ending March 2022 evaluation report

Updated 2 February 2023

Applies to England and Wales

Executive summary

In 2019, the Home Office announced that 18 police force areas (PFAs) would receive funding to establish (or build upon existing) Violence Reduction Units (VRUs) as part of the Serious Violence Fund. The aim of VRUs is to lead and coordinate a preventative, whole-system approach to violence reduction, which comprises:

• multi-agency working
• data sharing and analysis
• engaging young people and communities
• commissioning (and delivering) evidence-based interventions.

The Serious Violence Fund also covers Grip (formerly Surge) activity, which is enforcement focused. Combined, VRU and Grip funding represents a total investment of £254 million across the years (1 April to 31 March) 2019 to 2020 and 2021 to 2022 to tackle Serious Violence (SV)^{[footnote 1]}. The Home Office originally selected 18 areas and allocated funding amounts based on levels of SV between the years ending March 2016 to March 2018.

The Home Office commissioned Ecorys UK, Ipsos UK, the University of Hull and the University of Exeter to evaluate VRUs in their third year of operation (April 2021 to March 2022). The aims of the evaluation were to:

• estimate the impact of VRUs on violence
• assess the progress made by VRUs towards a whole-system approach

The mixed-methods evaluation included desk-based research, qualitative research with a range of stakeholders in each VRU, and quasi-experimental designs.

Key findings

There were no statistically significant impacts on the primary SV outcomes of hospital admissions for sharp object violent injury or homicides. However, there was a statistically significant reduction in police recorded violence without injury offences, which was a secondary outcome of interest. Since funding began, an estimated 136,000 (or 243 per 100,000 persons) violence without injury offences had been prevented in funded areas. Based on these offences avoided, a return on investment of £4.10 for every £1 of SV funding (VRU and Grip combined) was estimated^{[footnote 2]}.

Whilst not statistically significant, there were encouraging indications of reductions in homicides and hospital admissions resulting from any violent injury (for example, not just sharp object). Furthermore, analysis focusing on just violence ‘hot spots’ (within VRU areas) also indicated a potential impact on police recorded violence with injury offences.

Overall, VRUs made good progress towards a whole-systems approach. Building on progress made in previous years, VRUs showed signs of maturing and becoming embedded in local responses to prevent violence.

Key overarching findings against each element of the whole-systems approach included:

• multi-agency working:

  • as well as building on progress made in previous years on the development, leadership and co-ordination of a shared response to violence, the benefits of a multi-agency response to violence were emerging; this included reduced duplication and partners being better able to respond to violence
  • particular areas of progress in year ending March 2022 related to VRUs becoming more established and recognised by partners, and strengthening links between partners; non-financial support from VRUs, such as staff training, was valued and considered more sustainable (than financial support alone)
  • progress was more limited in some VRUs where certain partners (typically, health and voluntary sector partners, where capacity to engage may have been more limited) were less engaged and/or owing to the complex landscape of services, where some duplication was still evident

• data sharing and analysis:

  • VRUs continued to make progress in accessing a range of relevant data and using this to further their understanding of violence and, increasingly, inform and target their response
  • particular areas of progress in year ending March 2022 included VRUs overcoming some of the challenges faced in previous years around accessing data and developing data outputs that provide more actionable intelligence
  • variation in progress typically related to the granularity of data accessed by VRUs; VRUs that could access data with the capacity to analyse sub-area / ‘hotspot’ or individual-level data were better able to target VRU activity and support; accessing individual-level data has proved difficult for many VRUs owing to its sensitivity (for example, health data)

• engaging young people and communities:

  • there was some progress towards effective engagement with young people and communities; progress was most evident in VRUs that had developed more formal structures and roles to embed their engagement work
  • particular areas of progress in year ending March 2022 included how most VRUs had increased and diversified their engagement activities, compared to progress in the previous year, which had been limited by coronavirus (COVID-19) restrictions
  • all VRUs recognised the value of young person and community engagement but accepted that this requires further development

• commissioning and delivering interventions:

  • there was progress towards both strategic and evidence-based commissioning; in year ending March 2022, VRUs’ approaches to the commissioning, and governance, of interventions matured and became more purposeful
  • particular areas of progress in year ending March 2022 included almost all VRUs engaging with partners to commission interventions based on identified needs, and subsequent effective targeting of these interventions
  • progress was more mixed in relation to VRUs contributing to the evidence base, commissioning interventions informed by evidence of effectiveness and maturing towards a model of proportionate outcomes-based commissioning; this reflected some disparities in monitoring and evaluation capacity between VRUs

The VRUs’ contribution to the progress made towards a whole-systems approach was substantial. Whilst some progress may have been made in the absence of the VRU, it would have been over a longer period and/or reduced scale.

Recommendations

Where feasible and appropriate (for example, within their specific context), key recommendations for VRUs include:

• continue to engage with a wide range of partners, in particular the Core Members from health and voluntary sector organisation (VSO) partners where engagement may again have been limited, to further the whole-systems approach
• linked to the above, work with partners to ensure their role within the response to violence is strengths-based, proportionate and manageable; this would help ensure meaningful engagement and added value from partners, including those with more limited capacity; this means VRUs should continue to focus on the quality of partnerships, which might require focusing on key partners initially and gradually involving others
• as far as possible, focus on providing non-financial support to partners to support and develop their capacity, which supports more sustainable and longer-term impacts
• aim for access to at least sub-area level data from key sources so that interventions and wider activity (for example, multi-agency responses to hot spots) can be targeted at areas/groups most in need; VRUs could facilitate this through sharing learning and tools (for example, standardised data-sharing protocols)
• ensure there is sufficient expert capacity available to, or within, the VRU to negotiate access to data and provide actionable insights from the analysis; this could include expanding data teams, working with external experts and/or learning from other VRUs
• ensure there is representation in the Core Membership from community leaders and representatives of groups/grassroots organisations that can provide valuable insights to (and experience of working with) the at-risk cohort
• focus on co-development with young people and communities to ensure that evidence-based interventions and VRU activities are tailored to local contexts and needs. consider drawing on the evidence base to inform engagement approaches so that they effectively engage different groups; where possible, consider a dedicated community engagement role to facilitate this
• ensure that updated Response Strategies clearly articulate how interventions meet identified needs (based on data), that the VRUs capture these in both their Theory of Change (ToC) and Evaluation Plan/Strategy, and that progress towards meeting these needs can be monitored; this will help ensure interventions are evidenced-based and developed to meet local needs
• ensure that intervention portfolios include interventions where there is strong evidence of effectiveness to maximise the impact of the VRU; this could include interventions detailed in the Youth Endowment Fund Toolkit (or similar) and local evaluation findings; where existing evidence is more limited, VRUs should ensure they conduct appropriate monitoring/evaluation to build the evidence base
• develop monitoring and evaluation strategies; this would improve the consistency, comparability and quality of evidence generated, which would provide learning for future commissioning decisions; where possible and appropriate, VRUs could seek to strengthen their evaluation capacity through increased collaboration with the What Works Centres and research bodies (for example, local universities)

Recommendations for the Home Office, which would support the above, include:

• exploring with the Department for Health and Social Care whether they can encourage and support health partners in VRU areas to increase engagement and share data; this should include clarifying what data can and cannot be shared under current/upcoming (for example, SV Duty) legislation and clearly communicating this with VRUs and partner agencies
• where good practice has been identified (for example, trauma-informed training, standardised data-sharing protocols), support the sharing of tools and resources for wider implementation
• increase support and guidance to VRUs to become more sustainable (for example, types of support provided to partners, bid writing and demonstrating value to partners)
• exploring if the Home Office can work with the NHS to provide (extracts of) national administrative data to VRUs directly (for example, from NHS Digital rather than through specific trusts) at a sub-area or individual level
• revisiting and increasing the policy focus on the expectations around community and young person engagement
• develop guidance and templates for effective VRU intervention monitoring and evaluation strategies, emphasising the importance of accurate data records

Report authors and acknowledgments

Report authors:

• Ecorys UK: Matthew Cutmore, Jo Llewellyn, Shona MacLeod
• Ipsos UK: Jessica Baxendale, Meera Craston, Catherine Fenton
• Professor Iain Brennan (University of Hull)
• Professor Mark Kelson (University of Exeter)

The report authors would like to thank the VRU teams, VRU partner agencies, young people and communities who facilitated and took part in the research. Thanks also to the Home Office team and the wider research teams at Ecorys UK and Ipsos UK for their support with the evaluation.

1. Introduction

This report presents findings from the evaluation of Violence Reduction Units (VRUs) in the year ending March 2022, the third consecutive year of Home Office funding. The Home Office has funded VRUs in 18 Police Force Areas (PFAs) since the financial year 2019 to 2020, primarily to provide leadership and strategic co-ordination of all relevant agencies to support a ‘whole-system’ approach to tackling Serious Violence (SV). The Home Office commissioned this VRU evaluation in July 2021. The report builds on previous years’ evaluations, which were conducted by the same research team comprising Ecorys UK, Ipsos UK, Professor Iain Brennan of the University of Hull and Professor Mark Kelson of the University of Exeter.

This chapter provides the policy background to and programme-level Theory of Change (ToC) for VRUs, the evaluation aims and an overview of the evaluation methodology.

1.1 Policy background

In the summer of 2019, the Home Office announced that 18 PFAs would receive funding to establish (or build upon existing) VRUs as part of the Serious Violence Fund announced by the Treasury earlier that year. The Home Office selected the areas by the levels of SV experienced between year ending March 2016 and year ending March 2018. The Home Office confirmed funding for a second year (April 2020 to March 2021) and then a third year (April 2021 to March 2022), and additional funding made available to VRUs in year ending March 2022 for Serious Violence Youth Intervention Programmes.

The core aim of a VRU is to provide leadership and strategic co-ordination of all relevant agencies, to support a whole-system approach to prevent SV.

Key elements of the whole-system approach, include^{[footnote 3]}:

• multi-agency working to enable an efficient and effective responses to violence
• accessing and sharing data to ensure the response targets those in need
• engaging with communities and young people to guide the response
• commissioning and delivering evidence-based interventions to effectively support those in need

Key (annual) outputs which inform each VRU’s response to violence include a Strategic Needs Assessment (SNA), which provide insights to local violence trends and the at-risk cohorts, and a Response Strategy (RS), which sets out how the VRU is seeking to prevent violence, in the context of the SNA.

The Home Office specified 3 primary outcomes against which it would measure VRUs:

1. Reduction in hospital admissions for assaults with a knife or sharp object and especially among victims aged under 25.
2. Reduction in knife-enabled SV and especially among victims aged under 25.
3. Reduction in all non-domestic homicides and especially among victims aged under 25 involving knives.

The Serious Violence Fund includes the new ‘Grip’ (formerly Surge) programme. All 18 PFAs received Grip funding to roll out targeted, analysis-driven policing operations in the highest crime hotspots, rather than the more preventative focus of VRUs. The Serious Violence Fund represents a total investment of £254 million over 3 years.

1.2 Programme-Level Theory of Change

Figure 1.1 shows the VRU programme-level ToC. The ToC sets out the intended outcomes and impacts of VRUs and the causal pathways (indicated by arrows) to reach these through the inputs, activities and outputs. Activities and their anticipated outputs and outcomes have been structured, horizontally, by the key elements of the whole-systems approach:

• multi-agency working
• data sharing and analysis
• engagement with young people and communities
• interventions

Whilst each element comprises distinct activities, outputs and outcomes, they are complementary and collectively contribute to the anticipated (longer-term) impacts of VRUs. In particular, multi-agency working is considered a key facilitator of the other elements.

Figure 1.1: Programme-level Theory of Change for VRUs

1.3 Evaluation aims

The Home Office commissioned the evaluation team for a third consecutive year to conduct an impact and process evaluation of VRUs. The broad aims of the 2021 to 2022 evaluation were to:

• estimate the overall impact of SV funding (such as, VRUs and Grip activity) on violence outcomes and, as far as possible, understand the specific contribution of VRUs to reductions in violence
• assess the progress VRUs are making towards the different elements of the whole-systems approach to violence reduction, and consider the impact of VRUs on wider (more qualitative) outcomes related to embedding a whole-systems approach to violence reduction
• explore how VRUs in receipt of additional funding for new Serious Violence Youth Intervention Programmes commission and deliver these interventions

A programme-level ToC (section 1.2) and a linked evaluation framework (section 1.4) underpinned the evaluation.

1.4 Evaluation methodology

The following sections summarise the evaluation methodology. Annex B provides further technical detail (including limitations) regarding the quasi-experimental design (QED).

Estimating the impact on violence reduction

The evaluation team estimated the impact of SV funding for the following outcomes:

• primary outcome measures (linked to and where data was available for the Home Office specified outcomes):
  • hospital admissions for sharp object violent injury^{[footnote 4]}
  • (non-domestic) homicides

• secondary outcome measures, where, recognising the whole-systems and preventative approach of VRUs (see programme-level ToC), the team deemed it reasonable to explore potential impacts on the following wider forms of violence:

  • hospital admissions for any violent injury
  • police recorded violence with injury
  • police recorded violence without injury (which are mainly assaults not resulting in injury) (ONS, 2022)
  • police recorded possession of weapons offences (as an indicator of weapon carrying)

Data for all outcomes covered all police force areas (PFAs) and were available up to December 2021. The exceptions to this were Greater Manchester, where police recorded violence and weapons data were not available, and homicides data that were only available up to September 2021.

Whilst some caution is advised around interpreting police recorded outcomes where an increase could reflect improvements in reporting/detection and recording (see Annex B), their inclusion aligns to the preventative focus of VRUs and wider SV funding. This was supported by existing evidence of hotspot policing (a key Grip activity) and preventative interventions (commonly funded by VRUs) demonstrating impacts across a broad range of offences (Braga et al., 2019; Gaffney et al., 2021a and 2021b).

The primary QED approach to estimate impacts was synthetic control methods (SCMs). SCMs compare trends in outcomes following an intervention in treated areas (SV funded) to a synthetic control group. The synthetic control group is a weighted average of comparator areas (non-SV funded). Comparator areas that are most similar to treated areas on pre-intervention outcome trends receive a higher weighting. This approach, alongside population adjusting outcomes of interest and, where possible, focusing on smaller (than PFA) more comparable areas, allowed for comparisons between SV-funded and non-SV-funded areas that had similar outcomes trajectories (prior to funding), which were more appropriate than simple comparisons (without weighting or adjustment).

Exploratory analysis of violence ‘hot spots’, alongside qualitative insights from VRU stakeholders (see next section ‘Assessing VRU progress and wider outcomes’), were used to understand the specific contribution of VRUs towards violence reduction.

Assessing VRU progress and wider outcomes

To assess the progress of VRUs towards a whole-systems approach, and consider wider VRU outcomes, the team used both a process evaluation and theory-based impact evaluation.

For the theory-based impact evaluation, the team used contribution analysis. This involved developing and testing contribution hypotheses for each element of the whole-systems approach based on the programme-level ToC. Recognising VRUs are a longer-term approach to prevent violence, and to better understand VRU progress, the contribution hypotheses were broken-down into sequential stages:

Multi-agency working:

Stage 1: VRUs engage with relevant partners to develop a shared response to violence.

Key considerations for this stage included the extent to which partners were informing the direction of the VRU, a range of partners were (meaningfully) engaging with the VRU, and partner expertise and resources were being utilised by the VRU.

Stage 2: VRUs lead and coordinate the shared response to violence.

Key considerations here included the extent to which partners viewed the VRU as leading the multi-agency response to violence, and partners had clear roles in the response to violence, which they were supported (by the VRU) to take ownership of.

Stage 3: A more efficient and effective multi-agency response to violence.

This included consideration for the extent to which the multi-agency approach led by the VRU created efficiencies (for example, reduced duplication), the multi-agency response was perceived as effective, and the multi-agency response was considered sustainable.

Data sharing and analysis:

Stage 1: VRUs access and analyse data from relevant partners and agencies.

Key considerations for this stage included the extent to which partners were sharing the data (at the required level of granularity and quality to inform the VRUs response), and the VRU was working with partners to improve the quality and granularity of data shared.

Stage 2: VRUs identify and better understand trends in violence and the at-risk cohorts (and hotspots).

Key considerations here included the extent to which analysis provided the VRU and partners with a better understanding of violence and its drivers, and analysis had identified at-risk cohorts and/or areas.

Stage 3: VRUs’ response to violence reflects and is targeted towards the groups/areas that are most in need.

This included consideration for the extent to which analysis informed the VRUs response (for example, were those identified as at risk subsequently targeted with support) and informed the work/direction of partners, and the focus of the VRUs’ response to violence changed/evolved based on analysis.

Young people and community engagement:

Stage 1: VRUs engage young people and communities to inform the response to violence.

Key considerations for this stage included the extent to which the VRU was (effectively) engaging with those (most) affected by violence, and insights from engagement activity had improved the VRUs’ understanding of local young people and communities and subsequently informed the response to violence.

Stage 2: As a result of engagement activity, young people’s and communities’ perceptions of the VRU (and partners) improve.

Key considerations here included the extent to which young people and communities were aware of, and felt listened to by, the VRU.

Stage 3: VRU activities are co-developed with and informed by young people and communities.

This included consideration for the extent to which young people and communities co-designed VRU interventions and informed wider activities, and VRU interventions/activities were perceived as more effective because of this co-development.

VRU interventions:

Stage 1: Interventions are strategically commissioned and evidence-based.

Key considerations for this stage included the extent to which VRUs worked with partners to commission interventions strategically, interventions were commissioned based on identified needs and existing evidence of their effectiveness, and VRUs were contributing to the evidence base through evaluation and monitoring.

Stage 2: Interventions are appropriately targeted and effective at supporting those most in need.

Key considerations here included the extent to which interventions were targeted at, and effectively engaging individuals/areas identified as most in need, and interventions were perceived as effective.

Stage 3: Interventions lead to the intended outcomes for young people and communities.

This included consideration for the extent to which intended immediate/ shorter-term outcomes were being realised, and the ultimate/longer-term outcomes of reduced violence were materialising.

The team drew on a range of evidence to assess the VRUs’ progress within and across stages. The assessment focused on evidence of progress being delivered and whether stakeholders deemed this effective. Importantly, the assessment considered the different starting positions of VRUs.

Progress against each stage is reported at a programme-level (the general progress across all VRUs) using the terms defined in Table 1.1.

Table 1.1: Progress assessment terms

Levels of progress	Definition
Good	Broadly consistent evidence of (continued) progress being delivered, and this being perceived as effective
Some	Some evidence of (continued) progress, and this being perceived as effective
Limited	Limited evidence of (continued) progress, and/or this being perceived as effective

Levels of progress evident in (VRU coverage)	Definition
All	All 18 VRUs
Almost all/ all but a few	15 to 17 VRUs
Most	9 to 14 VRUs
Several	4 to 8 VRUs
Few	1 to 3 VRUs

Similar definitions were used when reporting progress across stages. For example, broadly consistent evidence showing progress from one stage to the next was considered ‘good’.

The process evaluation and contribution analysis utilised the following sources of evidence:

• documentary review: key documents for each VRU including year ending March 2022 funding applications, Strategic Needs Assessments, Response Strategies, and quarterly monitoring reports
• VRU-level ToCs: which were developed based on the documentary review
• qualitative research with stakeholders: a programme of in-depth consultations in all 18 VRU areas:
  • 72 interviews with VRU leads, strategic VRU stakeholders and external partners
  • 103 interviews with operational and analytical staff
  • 48 interviews with frontline staff (for example, delivering VRU-funded interventions)
  • interviews or focus group discussions with 50 young people supported by VRU-funded interventions, and 6 interviews with parents/carers of those young people
  • 23 consultations with young people and community representatives involved in developing the activities or strategic direction of the VRU

The evaluation team applied a purposive sampling strategy to the overall population of professional VRU stakeholders. The team requested a longlist of stakeholders from each VRU to ensure a range of coverage across organisations and roles of professionals. Young people involved with different types of interventions, and community members, were both purposively sampled and snowball sampled. The team then sampled the longlists of potential participants into shortlists to invite individuals to participate in the research.

The team transcribed the audio data collected, combined them with researcher notes, and sorted them into an analytical framework ready for analysis, alongside wider evidence and further data sources (such as VRU documents and monitoring data). All data at the individual VRU-level were analysed and triangulated to carry out assessments of progress against the contribution hypotheses and stages. The team used these stages to assess progress at the whole programme level.

To ensure the analysis was both inductive and deductive, the team analysed the qualitative data from interviews and focus groups ‘horizontally’ across the whole programme. This complemented the analysis mentioned in the paragraph above, which analysed ‘vertically’ through an entire individual VRU case study. The programme-wide analysis also allowed the team to find themes that may not emerge from a single case study alone, and therefore not emerge as a pattern in the data if only carrying out analysis at the case study level.

The team used NVivo qualitative data analysis software for the secure storage and analysis of the vast dataset. The analysis was halfway between inductive and deductive, through using a coding framework aligned to the ToC and contribution stages, as well as ground-up thematic content analysis (Anderson, 2007), which stemmed from grounded theory analysis approaches (Walker & Myrick, 2006).

Serious Youth Violence Interventions programme

A distinct yet integrated strand of the evaluation was to understand the implementation of the ‘Serious Youth Violence Interventions Programme’, which included a further document review and focused interviews with 36 commissioning and delivery staff.

1.5 Limitations

The evaluation was designed to ensure as robust and comprehensive approach as possible, combining multiple methods of data collection and analysis to provide a thorough investigation of how, why and in what context differing levels of success have been achieved within VRUs and at the programme-level.

However, there are some limitations to consider when reading the evaluation findings in this report. Key limitations to the data collection and analysis were:

Identification, selection and engagement of young people in the research:

The Omicron variant was dominant when the evaluation team was ready to carry out face-to-face fieldwork with young people. Because of the ongoing COVID-19 risks, some interventions were reluctant to grant access to the young people they worked with, so the evaluation team had to resample and begin the contact chain with another intervention, which took notable time and resource. Further, whilst research teams took all practical steps to avoid potential biases, there were limitations in their ability to target and sample research participants (for example, they could only interview young people engaged by projects).

Difficulties accessing frontline stakeholders, largely because of COVID-19:

Changing restrictions related to the pandemic reduced the availability of frontline staff to participate in the research. The proportion of frontline staff was lower than anticipated, which limited the triangulation of their views with strategic and operational stakeholders.

Timeframe required for observing outcomes and impacts:

The whole-system approach adopted by VRUs was intended to be long term. Longer-term outcomes and impacts as outlined in the programme-level ToC are only likely to be fully observable beyond the timescales of this evaluation.

Variation of monitoring and evaluation data on VRU-funded interventions:

The quality and coverage of available data on VRU-funded interventions varied greatly, affecting the ability to estimate the scale of work with young people and the possibility of assessing outcomes. The timescales of local evaluations (commissioned directly by VRUs to, typically, evaluate funded interventions) have rarely fitted with those programme evaluations, which limited the ability to use this data.

The perceptions of VRU stakeholders primarily informed assessments of progress:

Whilst the team interviewed multiple stakeholders, internal and external to the VRUs, to gather a range of perspectives to allow for a balanced assessment, it was not always possible to validate these against other sources of data. Notably, the aforementioned limitations of VRU data on funded interventions meant the evaluation was reliant on stakeholder perceptions (including those of young people) to (cautiously) assess the effectiveness of interventions.

Understandings of communities engaged, and community safety, remain limited:

There was limited community engagement activity across some VRUs and this influenced the data the team could collect in those areas relating to communities and perceptions of safety. Furthermore, community safety data collection at-scale is beyond the scope and remit of this study.

VRUs generally adopting/moving towards a ‘hybrid’ model:

As VRUs have evolved, most now have a central VRU team (covering the whole PFA) with some more localised elements (for example, devolving funding to Community Safety Partnerships (CSPs) and/or having local authority co-ordinator roles). This limited how much the team could assess a centralised model versus ‘hub and spoke’ (where there is a VRU ‘spoke’ in each local authority) model. Furthermore, only 3 VRUs had a ‘hub and spoke’ model, which also limited the assessment.

1.6 About this report

The report contains the following chapters:

• chapter 2 reports the results from the QED, assessing the impact VRUs have had on reducing SV
• chapters 3, 4, 5 and 6 provide an assessment of VRUs’ progress and contribution towards each element of the whole-systems approach. Each chapter starts with an assessment of progress across stages (see Figure 1.1) and the evidence of progress within each stage; approaches that appeared to be effective (facilitating progress) are then discussed; each chapter concludes with an overall judgement of VRUs’ contribution to the progress made
• chapter 7 presents conclusions on the extent to which VRUs have contributed to violence reduction and wider aims, recommendations arising, and areas for further investigation

2. Impact on violence

This chapter presents results from the quasi-experimental impact evaluation, which provided quantitative estimates of the impact of SV funding on violence outcomes. Results are then discussed alongside findings from the theory-based evaluation, where the specific contribution of VRUs towards violence reduction was explored.

Chapter 1 summarises the methodological approach, and Annex B provides additional methodological detail and outputs from the quasi-experimental impact evaluation.

2.1 Impact on violence outcomes

At the whole programme level (across all VRUs), there was no statistically significant impact on the primary outcomes of hospital admissions for sharp object violent injury or homicide rates. However, there was a general trend towards a relative reduction in (all and non-domestic) homicide rates compared to the synthetic control group (the primary QED used). Furthermore, previous research suggests that statistically significant impacts on the primary outcomes may take longer to materialise. For example, whilst research into the Cardiff Model demonstrated an impact on hospital admissions emerging a year after its implementation, there was a substantial setup period (33 months) to develop the model and required data sharing (Florence et al., 2011). It will be important for future research to continue to monitor these outcome trends as VRUs become more developed and embedded in local responses to violence.

Regarding the secondary outcomes of interest, there were statistically significant reductions in police recorded violence without injury and possession of weapon offence rates. Across all VRUs (apart from Greater Manchester, where data was unavailable), it was estimated that 136,257 (or 243 per 100,000 persons) violence without injury offences had been prevented. Although not statistically significant, there were indications of a relative reduction in hospital admission rates for any violent injury.

Table 2.1 details the programme-level cumulative impact estimates for all outcomes of interest following 2 years and 9 months (6 months for homicides) of SV funding (the ‘treatment period’, which covered from April 2019 to December 2021). The team analysed all outcomes as quarterly rates per 100,000 persons using local authority (or broadly equivalent CSP) level data. The exception to this was homicides, which was quarterly rates per 1 million persons and PFA data.

In any statistics model, there is some uncertainty around the (impact) estimates due to variability in the data. Confidence intervals (CI) provide a range of plausible values for the sample mean that is consistent with that observed. If the CI range includes zero (where the true value could be positive or negative), the team did not consider the impact estimate statistically significant (at the 95% level). Statistically significant estimates have an asterisk in the table.

Table 2.1: Programme-level impact estimates- synthetic control group analysis

Outcomes	Estimate	Lower CI	Upper CI
Primary:
Hospital admissions for sharp object violent injury	-0.02	-0.10	0.07
Homicides (per 1 million persons)	-2.84	-6.67	1.05
Non-domestic homicides (per 1 million persons)	-3.04	-7.35	1.19
Secondary:
Hospital admissions for any violent injury	-0.36	-0.79	0.08
Police recorded violence with injury	-2.05	-125.16	76.11
Police recorded violence without injury	-243.22*	-541.82	-51.58
Police recorded possession of weapons offences	-15.63*	-28.1	-0.59

All primary and secondary outcomes detailed in Table 2.1 show a reduction relative to the synthetic control group. Outcomes generally moving in the same direction provide additional validation of the results. However, it is important to consider the following:

• only the secondary outcomes of police recorded violence without injury and possession of weapon offences (considered an indicator for weapon carrying) were statistically significant (at the 95% level)
• some caution is advised for the outcome of police recorded possession of weapons offences, where an increase can result from increased detection (for example, linked to Grip activity), which can make this outcome difficult to interpret; furthermore, if a specific offence includes violent injury or another (more serious offence) alongside possession of weapons, it is likely only the former would be recorded (as the ‘primary’ offence); thus, there is potential for under recording of weapons possession
• the reduction in police recorded violence with injury was relatively small

The following graphs and discussion show additional outputs from the synthetic control group analysis for the primary outcomes and police recorded violence without injury offences (secondary outcome where there was a statistically significant reduction). Annex B provides outputs for the other outcomes.

Figure 2.1 shows the average quarterly hospital admissions for sharp object violent injury rates (per 100,000 persons) for VRU areas and the synthetic control group constructed. Relative to the (unweighted) average hospital admissions for sharp object violent injury rates in non-VRU areas, the synthetic control group provides a better counterfactual. Prior to SV funding, hospital admissions for sharp object violent injury rates were increasing in VRU and non-VRU areas. Following SV funding, there appeared to be a reduction in VRU and non-VRU areas. The increase in quarter 2, 2021 (July to September) corresponds to restrictions lifting following the first national COVID-19 lockdown. Differences following SV funding in hospital admissions for sharp object violent injury rates between VRU areas and the synthetic control group (which would indicate an impact) appear limited.

Figure 2.1: Hospital admission rates for sharp object violence (fiscal quarters) – synthetic control group analysis

Figure 2.2 shows the average quarterly homicide rate (per 1 million persons) for VRU areas and the synthetic control group constructed. Whilst relative to the (unweighted) average homicide rates in non-VRU areas, the synthetic control group provides a better counterfactual; there was some divergence in trends in 2018. Prior to SV funding, homicides appeared to be increasing in VRU areas. Following SV funding, the trend in VRU areas appears to level-off and tended to be lower than in the synthetic control group (indicating a potential impact, although not statistically significant).

Figure 2.2: Homicides (fiscal quarters) – synthetic control group analysis

Figure 2.3 shows the average quarterly violence without injury (per 100,000 persons) for VRU areas and the synthetic control group constructed. Prior to SV funding, violence without injury offences were increasing. Following SV funding, there is a clear divergence between the average violence without injury rate in VRU areas and the synthetic control group, which indicates a potential impact (this was statistically significant).

Figure 2.3: Violence without injury (fiscal quarters) – synthetic control group analysis

2.1.1 Cost benefit analysis

The team conducted cost-benefit analysis using the results from the police recorded crime (PRC) synthetic control group analysis. This is focused on the secondary outcome of violence without injury, as it was one of the 2 statistically significant results.

The cost-benefit analysis did not include the other statistically significant outcome for possession of weapon offences. This is because there was an absence of published financial benefits, and the aforementioned limitations of it as an outcome measure.

Table 2.2 details the total estimated offences avoided, which included applying multipliers to account for under reporting (see Heeks et al., 2018) and to reflect the total population in SV-funded areas and associated financial benefits. Based on the central estimate from the synthetic control group analysis, the total estimated benefits were £883 million. As well as this, the table provides estimates based on the lower and upper CIs.

Table 2.2: Estimated reductions in violence without injury and associated financial benefits

Estimate	Estimated reduction (per 100,000)	Multiplier (1.5)	VRU population (100,000s)	Total estimated reduction	£ Benefit (£6,480 per offence)
Central	-243.22	-364.83	373.48	-136,256.71	£882,943,470
Lower	-541.82	-812.73	373.48	-303,538.40	£1,966,928,835
Upper	-51.58	-77.37	373.48	-28,896.15	£187,247,036

After accounting for impact estimates only covering up to December 2021 (rather than the entire year up to March 2022) and the absence of Greater Manchester from the data, around £215 million has been invested in the SV funding since its inception in 2019. When considering the estimated benefits, the cost-benefit ratio (for the central estimate) was 4.1, meaning that for every £1 invested in SV activity, there was an estimated return of £4.10.

Table 2.3: Serious Violence costs, benefits and cost-benefit ratios

Estimate	£ Total SV funding	£ Benefit (£6,480 per offence)	Cost-benefit ratio
Central	£215,094,869	£882,943,470	4.10
Lower	£215,094,869	£1,966,928,835	9.14
Upper	£215,094,869	£187,247,036	0.87

2.1.2 Impact on violence hot spots

To understand the impact of SV funding in areas most affected by violence, the team undertook exploratory synthetic control group analysis using Lower Super Output Area (LSOAs)^{[footnote 5]} level police recorded violence data. The areas most affected by violence (‘hot spots’) were defined as LSOAs with the highest levels (top 5% within each PFA) of violence prior to SV funding. Recognising the analysis was exploratory, and that reliable LSOA-level data were only available for 9 SV-funded areas^{[footnote 6]}, some caution is advised with the results from this analysis.

The team intended the analysis to primarily provide insights into the impact of Grip (and its predecessor, Surge), where there was an explicit focus on hot spot policing. Whilst VRUs were expected to target activity at individuals and areas most in need, the team anticipated their impact to be PFA-wide.

The analysis estimated a total reduction (over 2 years and 9 months) of 19,420 (95% CI [22,591, -16,219]) violence with injury offences in SV-funded hotspots, relative to the synthetic control group. However, some caution is advised with these findings owing to data limitations and greater apparent variability at the LSOA-level of the impact of COVID-19 restrictions on violence trends. The latter requires further exploration when more recent data (less affected by COVID-19 restrictions) becomes available.

Whilst not directly comparable, the analysis of hot spots indicates SV funding (specifically, Grip activity) likely contributed to reductions in violence with injury offences, as well as the reductions in violence without injury offences estimated at the whole programme level (see section 2.1).

2.1.3 Exploring differences in impact and approaches

The team undertook meta-regression to explore whether different levels of SV funding or approaches on how the funding was used could explain differences in impacts between SV-funded areas. This involved estimating impacts on violence for each VRU using synthetic control groups and then treating these as individual ‘studies’ in the meta-regression. A range of explanatory variables related to VRU composition and structure, and funding levels/focus were tested.

Generally, differences in VRU-level impact estimates were not statistically significant from one another (the CIs were overlapping). The meta-regression revealed a potential association between levels of funding and hospital admissions with any violence. Impact estimates for VRUs with higher levels of funding (in the top funding tier) were more likely to indicate a reduction. Whilst statistically significant, all these effects were relatively small. These should form the basis for ongoing scrutiny in future evaluations.

2.2 VRU contribution towards violence reduction

The impact evaluation indicated a reduction in violence in SV-funded areas relative to non-funded areas. Whilst reductions for homicides and hospital admissions (SV) were not statistically significant, the direction of the former and statistically significant reductions in PRC, were encouraging within the analysis timeframes. Furthermore, analysis of ‘hot spots’ suggested that Grip and VRU activity were likely both contributing towards the reductions in violence observed.

To explore the contribution of VRUs further, the team consulted strategic and operational stakeholders in each VRU.

There was consensus across stakeholders that their VRUs were contributing to local efforts to reduce violence. This contribution was most evident in the VRUs’ multi-agency response to violence, which had aligned local partners’ aims and actions. Particular contributions in year ending March 2022 related to increased alignment between VRUs and Grip activity, and partnership working maturing and becoming more embedded. Stakeholders also referenced the VRU contribution of data and evidence sharing, which had enabled a more informed and targeted response, and the additional resource provided by VRUs enabling support to reach those at risk. Details of VRUs’ approaches to multi-agency working, data sharing and interventions are in subsequent chapters.

Stakeholders felt the work of the VRU was complementing, rather than duplicating, existing work. This had become more evident for some VRUs in Year 3 (year ending March 2022). Whilst some duplication was still reported, for example, owing to multiple funding streams to address similar underlying issues, the VRUs provided a forum to map and, as far as possible, align activity.

Generally, stakeholders were somewhat hesitant about attributing reductions in violence across their area directly to the VRU. Whilst they felt the VRU was contributing, it was difficult to disentangle impacts owing to wider linked activity (for example, Grip) and external factors (for example, COVID-19 restrictions), which impacted on violence trends. Whilst many stakeholders were still of the view that it was too soon to say whether the longer-term whole-systems approach of VRUs was reducing violence, the role and value of the VRU had become clearer in year ending March 2022. These timelines around impact align to evidence from the Scottish VRU, which was established in 2005. In Scotland, SV reportedly halved, but over an 8-year period^{[footnote 7]}. Some stakeholders argued that reductions at this stage were more likely because of Grip activity, where the impact is more immediate.

3. Multi-agency working

This chapter provides an assessment of VRUs’ progress and contribution towards effective multi-agency working. Multi-agency working is a core element of the whole-systems approach to violence prevention and can be considered a foundation for other elements.

Aligned to the programme-level ToC, progress towards effective multi-agency is considered against the following (connected/sequential) stages:

1. VRUs engage with relevant partners to lead the development of a shared response to violence.
2. VRUs lead and coordinate the response to violence, where ownership is shared with partners, which ultimately leads to,
3. A more efficient and effective multi-agency approach.

The assessment, which was primarily informed through the triangulation of interviews with multiple stakeholders and review of supporting documentation, covers overall progress since year ending March 2020 (the first year of VRU funding) but particular areas of progress in year ending March 2022 (that is, since the previous evaluation) are highlighted. The progress assessment is followed with discussions of VRU approaches to multi-agency working, which facilitated progress.

3.1 Progress towards multi-agency working

Overall progress towards multi-agency working has been good. As well as building on progress made in previous years, the benefits of a more efficient and effective multi-agency response to violence were emerging. This included reduced duplication and partners being better able to respond to violence. Progress was more limited in some VRUs, where certain partners were less engaged.

In year ending March 2020, VRUs focused on bringing together key partners to develop a shared response to violence in the form of an initial RS, which included establishing VRU structures and membership. They built on this initial progress in year ending March 2021, where VRUs focused more on the co-ordination of the response to violence by mapping and starting to align partners’ strategic aims and existing multi-agency structures. This provided the foundation for progress in year ending March 2022, where VRUs were increasingly perceived as leading and coordinating the response to violence, and there was evidence of more efficient and effective multi-agency working.

Particular areas of progress in year ending March 2022 related to VRUs becoming more established and recognised by partners, which was evident through increased and more focused engagement from partners. Links between partners were also strengthening because of the work of the VRU.

Progress against each stage of multi-agency working is discussed in the sections that follow. Evidence relating to progress was generally strong (at least broadly consistent within stakeholder groups).

3.1.1 Development of a shared responses to violence

The Home Office expected all VRUs to engage with relevant partners to develop a shared response to violence, which draws on partner expertise and resources. This engagement ensures partners buy into the shared response to violence.

Overall, almost all VRUs demonstrated good levels of progress in the development of a shared response to violence. Progress was more limited in some VRUs, where it was less clear how partner resources were being utilised beyond informing the strategic direction of the VRU.

It was evident across all VRUs that partners were informing the strategic direction of the VRU. This was demonstrated by VRUs consulting partners when developing their Response Strategies, and through more regular strategic meetings with Core Members (senior representatives from key partner organisations).

In year ending March 2022, VRUs had made progress with the range and, in some VRUs, where this had been a problem in previous years, the seniority of Core Members engaged. The former included increased representation of VSOs. The latter was perceived as substantial progress in affected VRUs, as Core Members needed to be able to influence the direction of the partner organisation they represented. The team attributed this progress to an increased recognition of VRUs amongst partners. Several VRUs still noted difficulties in engaging some key partners, such as health and education, where competing priorities and demands were a factor.

Progress was also evident in many VRUs around more focused engagement of partners. For example, the establishment of sub-groups, where the most relevant partners could focus on particular issues and themes.

Beyond informing the strategic direction of the VRU, clearly partner expertise and resources were being utilised in the response to violence in most VRUs, demonstrated through partners providing expertise to specific workstreams and sharing data with the VRU (see Chapter 4). Progress appeared to be more limited in some VRUs, where partners were mainly just informing high-level strategy.

3.1.2 VRUs lead and coordinate the response to violence

The core function of VRUs was to lead, coordinate and support partners to share ownership of the local response to violence. The Home Office considered a shared response to violence, which harnesses the resource and expertise of partners, key to VRUs achieving their ultimate aim to prevent violence.

Overall, most VRUs had made good progress in leading and coordinating the response to violence and supporting partners to take ownership. Variation in progress typically related to the extent to which different partners were taking ownership of their role, and the extent to which partners perceived the VRU to be leading the response.

Across all but a few VRUs, partners perceived the VRU as leading and coordinating the response to violence. Several VRUs noted that this perception had been strengthened in year ending March 2022, as they became more embedded and established. Some VRUs also mentioned the anticipation of the Serious Violence Duty supporting an increased recognition of the VRU. Progress was more limited in a few VRUs, where either some partners still perceived existing partnerships (for example, Community Safety Partnerships) as ‘leads’, or the VRU was not effectively linked into existing key strategic multi-agency boards (for example, safeguarding boards).

Some VRUs saw themselves as more of a facilitator for a coordinated response to violence at a more local level, where they centrally supported/enabled partnership working within local authorities. Partners considered this as effective and appropriate to their specific local context.

Generally, partners across all VRUs understood their role and how this fitted within the wider response to violence. There was evidence across VRUs of partners (increasingly) taking ownership of their role, through increased partnership working and partners leading on workstreams (for example, education-focused workstreams, ‘task and finish’ workstreams/groups), with the support from the VRU. This was most evident for those considered key partners, such as the police, where VRU expectations around ownership were typically higher. However, it was evident that not all partners were taking ownership of their role. For example, health and voluntary sector organisation (VSO) partners in several VRUs had not taken as much ownership of the response to violence as other partners. VRUs noted that general levels of engagement and the short-term nature of funding were barriers to progress with these partners.

There was evidence of partners effectively being held to account by the VRU to deliver on their roles and responsibilities in the response to violence. Approaches included formal terms of reference with partners and/or a common understanding and trust between the VRU and partners to deliver on the shared response to violence.

3.1.3 A more efficient and effective multi-agency approach

As a result of the preceding stages, the team anticipated a more efficient, effective and sustainable multi-agency approach towards violence prevention. For context, this stage, which included organisational and professional culture change, was considered a longer-term goal for VRUs.

Overall, and with the above context in mind, most VRUs had made good progress in facilitating a more efficient and effective multi-agency approach. Progress was most evident in VRUs focusing on professional development and building relationships between partners, which were considered more sustainable elements of the multi-agency approach.

Across all VRUs, there was evidence of efficiencies being made. Typically, these were achieved through reduced duplication in partners’ work, which was attributed to VRUs mapping and consolidating activity. A few VRUs still noted some duplication owing to the complex and evolving landscape of services. For example, it was sometimes difficult to align multiple central government (for example, Department for Education) funding streams focused on similar issues. Progress towards an efficient multi-agency approach was most evident in VRUs with established multi-agency case management, including referral pathways and crisis management.

There was emerging evidence in almost all VRUs of a more effective multi-agency response. VRUs had shared data and intelligence, and best practice around ‘what works’ with partners to enable them to respond more effectively. Many VRUs had also supported (frontline) professionals with training. Combined, this led to progress in partners and professionals understanding their role within, and being better equipped to deliver, the response to violence.

Evidence of progress towards a sustainable approach to multi-agency working was more mixed. Whilst partners in some VRUs felt the relationships they had developed with other partners (through the VRU) were sustainable, others were of the view that the structure and resources provided by the VRU was required to sustain a multi-agency approach. Elements noted as sustainable included trauma-informed training provided to professionals^{[footnote 8]}, and supporting changes to partners’ longer-term strategies and policies. An example of the latter included the VRU supporting education partners to update local schools’ exclusions policies for students found in possession of a knife.

3.2 VRU approaches to multi-agency working

Key to effective multi-agency working are VRU structures, leadership and membership. VRU structures, typically, comprised 3 levels:

1. Governance to oversee and hold the VRU and partners accountable.
2. Strategic (Core Members) to develop a shared ownership of the response to violence.
3. Operational to deliver the response alongside partners.

Many VRUs noted the role of Core Members as a key facilitator to multi-agency working. Core Members should be able to provide expertise to inform the direction of the VRU, and be senior enough to influence change within their organisation to align to and support the aims of the VRU. Approaches to effectively engaging and sharing ownership with Core Members and the partner organisations they represented included:

• ensuring meetings were focused and well prepared to maximise their value; this included focusing on a specific issue or theme and sharing data/reports with partners to discuss; regarding the former, some VRUs had established sub-groups, often referred to as ‘task and finish groups’, which comprised select partners to focus on a specific issue or theme related to their area of work; VRUs often perceived this as more effective than meetings with larger groups and/or where the focus was more general
• where possible, consolidating (multi-agency) meetings to avoid duplication and manage burden on partners
• effective communications, including sharing data and through social media, to demonstrate the role and value of the VRU to wider partners; as VRUs became more established in year ending March 2022, and the benefits of the whole-systems approach began materialising more visibly, many reported increased engagement from existing and wider partners; several VRUs harnessed this in their communications with partners
• preparing partners for the Serious Violence Duty coming into effect
• making partners strand or workstream leads, where they could take ownership and were accountable to deliver specific actions

At a more operational level, teams needed to be able to facilitate and support multi-agency working. Effective approaches included:

• core teams, including dedicated partnership managers to facilitate and oversee partnership working, including at local authority level (where variability in partnership working between and within was often noted)
• appointing sector specialists, often seconded from partner organisations or sectors, with the expertise and relationships to develop partnership working; examples included education leads (see case study below), and a Department for Work and Pensions lead to ensure a joined-up approach on the VRUs’ and partners’ diversionary work

Case study: Example of effective multi-agency working through developing direct links with schools

VRU model: Hub and Spoke.

Context: The VRU area covers multiple local authorities, each with their own specific local challenges, infrastructure and partnerships. To enable violence reduction responses to be designed to meet these different local needs/priorities, the VRU has one central (strategic) hub, and then multiple spokes working in the individual local authority areas.

Core team: The structure of the core team varies across the spokes, depending on the scale of activity required or planned. Each spoke has a VRU co-ordinator.

In year ending March 2021, one of the VRU spokes piloted an education inclusion programme, which was scaled up to the other spokes in year ending March 2022. The programme involves having education inclusion professionals working in and with schools across the spoke areas to change their behaviour policies and approaches to reduce exclusions, and directly with young people identified as vulnerable to exploitation and SV. The VRU spoke tailored the scope of the activity to schools’ and young peoples’ needs, but generally the education inclusion professionals work in relation to various issues such as attendance, attainment and exclusions.

Stakeholders working on the project felt that the work had been effective in supporting schools’ response to violence, and this work had improved in year ending March 2022 due to the professionals becoming more embedded in schools. Stakeholders suggested that the programme has facilitated – and has been facilitated by – multi-agency working in different ways. The key benefits of having dedicated education inclusion professionals have included providing a direct link between VRUs, schools and other agencies (for example, children’s social care), which has supported information sharing and increased trust between schools and the VRU; and professionals being able to refer young people to wider VRU interventions/partner agencies if needed.

The VRUs provided a range of support to partners to facilitate shared ownership and delivery of the response to violence. More direct financial support included:

• access to VRU-funded interventions that partners would have otherwise had to develop/deliver themselves or would not have existed at all (that is, without the additional funding)
• devolving funding to partners and/or local areas; in one VRU, this had included the development of a ‘crisis fund’ that local authorities could access when responding to an incident; other VRUs had set money aside for grants that partners could bid into and encouraged collaborative bids
• often linked to the above, support with developing local action plans, which draw on VRU analysis and clearly set out how identified local needs will be addressed; as well as ensuring accountability, local action plans also helped develop relationships between partners within areas

Linking back to the previous section, whilst partners welcomed such financial support, it was less likely to be considered sustainable (in the absence of VRUs). However, its potential value in bringing partners together to develop relationships and work more collaboratively should be noted.

Effective non-financial support, where there was often a stronger sustainability argument, included:

• supporting partner organisations with bid writing
• providing data and analysis to increase partners’ understanding around violence and facilitate effective targeting of areas, cohorts and individuals (see Chapter 4)
• sharing best practice of ‘what works’ drawing on locally generated and wider evidence (see Chapter 6); this included toolkits for interventions and evaluations
• supporting a public health approach and greater recognition of violence within the partner organisation; this included working with partners to align their agendas and professional development; the latter typically involved trauma-informed training, which was considered highly effective in supporting (frontline) professionals to understand their role in the response to violence and upskill them to respond effectively; a case study for trauma-informed training is below

Case study: Example of effective multi-agency working through delivering trauma-informed training to partners

VRU model: Centralised.

Context: The VRU was established in a context of there being a historic lack of strategic co-ordination and inconsistent/patchy delivery of early intervention projects across the area. A key principle of the VRU is to work as a whole system and intervene early to prevent issues emerging. Developing a multi-agency workforce with the knowledge and skills to work in a trauma-informed way is central to achieving this principle.

Core team: The core team includes the VRU director, a programme manager, community engagement support, business support and public health staff.

As part of the RS (developed from the SNA), the local agencies in the VRU area committed to a trauma-informed approach. Whilst initially training was available for statutory agencies, in year ending March 2021, the VRU established an offer to extend this training and support to the Voluntary and Community Sector (VCS) organisations. The VRU recognised the importance of extending the offer to these organisations, given their key role in working with young people at risk of being impacted by SV. Through this offer, the VRU provides training on, and related to, trauma-informed approaches. In year ending March 2022, it provided specific training on:

• safeguarding
• restorative resilience supervision
• cultural competency
• information sharing

Interviewed strategic and operational stakeholders saw the training programme as one of the VRU’s key achievements of year ending March 2022. Beyond providing training, stakeholders thought the programme represented the VRU’s efforts to recognise VCS organisations as ‘equal partners’ in responding to SV. Building on the success of the training programme so far, the VRU intends to establish a minimum basic standard of training and offer more comprehensive continuous professional development for practitioners. It also hopes to focus on ensuring the practices learnt through the training are applied in a joined-up way across statutory and VCS organisations.

In terms of VRU models, progress in multi-agency working (and other elements of the whole-systems approach) appeared to be most evident in centralised models. However, it is important to consider that only 3 VRUs had a ‘hub and spoke’ model and most VRUs were considered ‘hybrid’ models (rather than purely centralised), which both limit this assessment. Furthermore, stakeholders did not perceive the ‘hub and spoke’ model as a barrier to progress.

3.3 VRU contribution to multi-agency working

The VRU contribution to multi-agency working was significant in most VRUs. It is unlikely this progress would have been made in the absence of the VRU. In other VRUs, the contribution had accelerated and facilitated progress. Whilst this progress may have still been made in the absence of the VRU, it would have likely been over a longer period and/or at reduced scale.

The main contribution had been the leadership and co-ordination provided by the VRU, which had brought partners together with a common aim. By bringing partners together, new relationships had been formed and strengthened. Even in VRUs where there were existing multi-agency structures in place, stakeholders considered the specific focus and additional resource of the VRU a valuable contribution. Many stakeholders noted that the contribution of VRUs was more tangible in year ending March 2022.

4. Data sharing and analysis

This chapter provides an assessment of VRUs’ progress and contribution towards effective data sharing and analysis. Data sharing and analysis is a core element of the whole-systems approach to violence prevention.

Aligned to the programme-level ToC, progress towards effective data-sharing analysis is considered against the following (connected/sequential) stages:

• VRUs access and analyse data from relevant partners and agencies
• VRUs better understand trends in violence and the at-risk cohorts and areas, which ultimately leads to:
  • a response to violence that reflects and is targeted at those in need

The assessment, which was primarily informed through the triangulation of interviews with multiple stakeholders and review of supporting documentation, covers overall progress since year ending March 2020 (the first year of VRU funding) but particular areas of progress in year ending March 2022 (that is, since the previous evaluation) are highlighted. The progress assessment is followed with discussions of VRU approaches to data sharing and analysis, which facilitated progress.

4.1 Progress towards effective data sharing and analysis

Overall progress towards effective data sharing and analysis has been good. VRUs are accessing a range of relevant data and using this to inform their response to violence. VRUs with access to granular data could make more progress in the targeting of activities and supporting partners.

In year ending March 2020, VRUs focused on collating and analysing available data to inform their SNAs, which provided initial insights into trends in violence and the at-risk cohorts. Building on this initial work, VRUs could adopt a more focused approach to accessing and analysing data in year ending March 2021 to further their understanding. Continued progress in year ending March 2022 facilitated an increasingly informed and targeted approach.

Particular areas of progress in year ending March 2022 included VRUs overcoming some of the challenges faced in previous years around accessing data (for example, hospitals) and developing data outputs that provide more actionable intelligence.

The following sections discuss the progress against each stage of data sharing and analysis. Evidence relating to progress was generally strong (at least broadly consistent within stakeholder groups).

4.1.1 VRUs access and analyse data

The Home Office expected all VRUs to work with partners to access and analyse a range of data. The drivers of violence and risk factors are complex, and data related to them spread across different agencies. VRUs have a key role in bringing this data together and supporting improvements in data quality.

Overall, most VRUs demonstrated good levels of progress in accessing key data related to violence and supporting improvements to its usability and quality. Progress was more limited with accessing more granular data from wider sources.

All VRUs were accessing a range of data to inform their response to violence. Table 4.1 details the number (and percentage) of VRUs where a data-sharing agreement (DSA) was in place (or in progress) for different data sources by level of granularity. Key observations were:

• generally, it was more common for VRUs to access data at the population (Level 1) or sub-area/hotspot (Level 2) levels as compared to the individual level (Level 3); this was particularly noticeable regarding data related to hospital admissions, ambulance incidents/call-outs, social care, education, urgent care and housing
• whilst access to individual-level data was relatively uncommon for most data sources, many VRUs accessed individual-level data related to crime/SV (10), and for a few VRUs, social care (4) and education (3)
• most VRUs were accessing data related to hospital admissions and ambulance incidents/call-outs; respectively, 13 and 12 VRUs were accessing this data at least at the population level; furthermore, 9 and 10 VRUs were accessing this data at the sub-area/hotspot level; no VRUs accessed individual-level data here
• just under half of the VRUs were accessing data at least at the population and sub-area/hotspot levels related to social care (9 and 8 VRUs respectively), education (8 and 7) and hospital attendances (8 and 7); access to individual-level data was less common for these data sources among VRUs also (4 VRUs for social care, 3 VRUs for education, 1 VRU for hospital attendance)

Table 4.1: VRU access to data by source and level

	DSAs in place or in progress for at least population level data (Level 1+)		DSAs in place or in progress for at least sub-area level data (Level 2+)		DSAs in place or in progress for individual-level data (Level 3)
Data source	VRUs	%	VRUs	%	VRUs	%
Crime / Serious Violence	18	100%	16	89%	10	56%
Hospital admissions	13	72%	9	50%	0	0%
Ambulance	12	67%	10	56%	0	0%
Social care	9	50%	8	44%	4	22%
Education	8	44%	7	39%	3	17%
Hospital attendance	8	44%	7	39%	1	6%
Housing	6	33%	5	28%	2	11%
Health	2	11%	2	11%	0	0%
Probation	2	11%	1	6%	1	6%
Youth offending	2	11%	2	11%	1	6%
Fire services	1	6%	1	6%	0	0%
Prison	1	6%	1	6%	1	6%
Socio-economic	1	6%	0	0%	0	0%
Youth services	1	6%	1	6%	1	6%

Note that Table 4.1 only considers data where it was indicated a DSA was required. Publicly accessible data was not included, and it is possible there are some datasets (for example, data held by the police) where VRUs did not require a DSA.

Some VRUs could only access more granular (for example, ward- or individual-level) data for some parts (for example, local authorities or hospitals) of/within their area. This highlights the complexity of data sharing across large geographic areas. Similarly, progress was evident in the granularity of data accessed, but this did not extend to all sources. For example, accessing individual-/family-level data around safeguarding (such as, social care), education or hospital data was a challenge for many VRUs.

VRUs were working with relevant partners to improve the quality and granularity of data provided. For example, VRUs were working with health partners to improve the usability and comparability of hospital data, where there was often variation in recording across the area covered by the VRU.

4.1.2 VRUs better understand trends in violence and the at-risk cohorts

A key role of the VRU is to undertake analysis to develop a better understanding of violence trends and its drivers and identify the at-risk cohorts (and/or areas) to target support towards.

Overall, almost all VRUs made good progress in both senses. The VRUs making the most progress were those that were utilising data in or near ‘real time’ and/or to identify individuals at potential risk. Whilst the latter is connected to the granularity of data accessed, all VRUs were focused on maximising the value of the data they did have access to.

Analysis undertaken by the VRUs led to better understanding of local violence trends and its drivers, which was shared with key partners. Key to this was triangulation across different data sources to understand patterns. The VRUs considered real-time (or at least regularly updated) data particularly valuable given how quickly trends and patterns can change, such as due to COVID-19 restrictions and changes in criminal gang activity.

Almost all VRUs were using data to identify at-risk cohorts and/or areas (hot spots), where, linking to the previous stage, at least sub-area level data was typically required. For several VRUs, this included identifying at-risk individuals using police and wider data sources. Partners that could access such information (for example, through dashboards) found this very valuable. Where identifying individuals was not possible (owing to the granularity of available data), VRUs still made progress in identifying at-risk cohorts and areas. This included working closely with police/Grip teams to identify hot spots through the triangulation of multiple data sources.

For the few VRUs where progress towards identifying the at-risk cohorts and/or areas was more limited, this was due to the granularity of data available and analytical capacity. However, it was still evident that these VRUs had developed a better understanding of the at-risk cohorts and shared this with partners.

4.1.3 Responses to violence reflects and targets those in need

As a result of the preceding stages, the VRUs’ and partners’ response to violence should reflect the needs of and target those identified as at risk. This is key for an evidence-based approach where resources are targeted to have maximum effect.

Overall, most VRUs had made good progress with analysis informing their response to violence and the work/direction of partners. Progress was most evident in VRUs using data for case management/referral pathways. Lack of availability of granular data limited the progress of some VRUs.

Data was informing VRUs’ response to violence in several ways. SNAs remained a key tool for informing the direction of VRUs. Place-based targeting (for example, where those at risk are concentrated) of interventions was common in almost all VRUs (again, requiring at least sub-area level data). Several VRUs had progressed even further with (multi-agency) case management/referral pathways for those identified as at risk, which was facilitated by access to individual-level data, analysis tools (for example, dashboards) and multi-agency meetings. Whilst all VRUs had used data to inform the types of interventions they commission, some were unable to target these at particular areas/cohorts as well as other VRUs due to only high-level data being available.

The analysis and data provided by VRUs was also informing the work of partners. This was particularly evident with the police, where many VRUs had worked closely with them to identify hot spots. Other evidence of progress included analysis highlighting gaps in provision and keeping partners updated with changes in violence trends. An example of the former involved the CSPs hiring additional youth workers based on needs identified by the VRU.

4.2 VRU approaches to data sharing and analysis

Approaches to accessing more granular data, which facilitated progress, typically involved VRUs harnessing their wider multi-agency function to work through barriers. Some VRUs had formalised this through the establishment of sub-boards where the VRU and relevant representatives from partner agencies came together to discuss data sharing. This was perceived as an effective approach. Furthermore, there was evidence of VRU Directors and Core Members using their seniority and connections to ‘unblock’ data sharing in some VRUs.

Different and more innovative approaches to data sharing had also developed. For example, some VRUs had developed (or built on existing) standardised data-sharing protocols, which partners opted into. This helped navigate the challenges and resources associated with drafting separate DSAs. This was sometimes facilitated through wider external expertise to develop data dashboards, where academic experts had already developed sharing protocols (in similar contexts / with other VRUs).

One of the larger VRUs was focusing on facilitating data sharing between agencies within the CSPs it covers, rather than tackle this centrally, which they considered unfeasible.

Some VRUs noted that partners often used different data collection systems and formats, which could prove challenging, so they were working to overcome this by supporting partners to standardise their data collection.

VRUs also considered the development of data dashboards and other data ‘products’ an effective approach to encourage wider partners to share data. Such data products demonstrated the usefulness of access to multiple data sources in informative and accessible format and, alongside overcoming partners’ risk-aversion, were consistent with wider research into data sharing.

The data sharing challenges outlined above were consistent with those detailed in literature (Goldacre et al., 2022), also demonstrating that these challenges are not unique to working in the violence prevention space. For example, focus groups and interviews with police and VRU analysts involved in the prevention of SV raised concerns about the quality of multi-agency data and the extent to which data sharing between services actually takes place (Bullock et al., 2022). Brophy et al. (2022) highlighted the need for senior level involvement to agree and provide clarity on what can and cannot be shared, standardising data collection/systems across police forces and partner agencies, to demonstrate the benefits of data sharing are key facilitators.

Several effective approaches were evident to understand violence trends, and identify the at-risk cohorts and areas. Key was regular analysis outputs and focused discussions about these. The latter allowed for wider intelligence from partners to be considered and refinements (including prompting action to address current gaps in the data) made to future analysis. Linked to this was ensuring there was analytical capacity within the VRU to provide regular and insightful analysis. Many VRUs attributed more recent progress in their data sharing and analysis work to moving from often sole/seconded analysts towards dedicated analysis teams or ‘data hubs’.

Close working with Grip teams was also a part of many VRUs’ approach to identifying hotspots, which provided the police with insights from data that they had not previously had access to. Innovative approaches to identifying at-risk individuals were evident in some VRUs, which included calculating risk scores based on indicators across different data sources, and multi-agency reviews following SV incidents.

To translate insights from data into the VRUs’ response, approaches included a multi-agency response to identified hotspots. In particular, this included working closely with the police to ensure a joined-up enforcement and prevention response. This approach had strengthened VRU and police relations. Where VRUs could identify at-risk individuals, case management had started to become part of the VRUs’ approach.

Case study: Using data to identify and target those at risk

Analysis of pre-existing and newly accessed data sources had allowed the VRU to identify and better understand trends in violence and the at-risk cohorts. Stakeholders reported the VRU had moved into a ‘new territory’ in enabling a ‘data-driven partner response’. Key to this has been the development of an internal dashboard (for the police) and a multi-agency dashboard – which were now acting as ‘building blocks’ to:

• identify individuals already causing significant harm or those most at risk of causing harm
• drive the operational and multi-agency response plan around individuals and cohorts of individuals identified

The internal dashboard, which is relatively more developed actively, has informed police operational activity, such as:

• enabling hotspot policing on the LSOA level
• running an anti-knife crime campaign in selected areas of the city using VRU data to target high-violence areas in which to broadcast the campaign
• enabling identification of at-risk individuals and using the classification of ‘3 cohorts’
• better understanding of individuals (victims or perpetrators, type of harms), highlighting exposure to risk factors, and showcasing evidence-based options for response around what interventions might work

4.3 VRU contribution to data sharing and analysis

The VRU contribution to data sharing and analysis was significant in most VRUs. Although possible, this progress was probably unlikely or would have taken much longer without the VRU. In other VRUs, the contribution had accelerated and facilitated progress.

The main contribution had been the dedicated resource provided by the VRU to facilitate data sharing and provide analysis to partners. Even in VRUs with existing data-sharing agreements, the VRU contribution was still recognised as bringing that data together with a view to prevent violence. VRUs also contributed to improvements in the quality and use of data and supporting partners with limited analytical capacity.

Reference was also made to the contribution of the seniority and motivation of VRU Directors and Core Members who could support and facilitate data sharing.

5. Community engagement

This chapter provides an assessment of the VRUs’ progress and contribution towards the effective engagement of young people and communities. This is one of the core elements of the whole-systems approach to violence prevention.

Aligned to the programme-level ToC, the chapter considers progress towards effective engagement with young people and communities against the following (connected/ sequential) stages:

1. VRUs have engaged with young people and communities to inform the VRUs response to violence.
2. Through this engagement, young people and communities have improved perceptions of VRUs, partners and interventions.
3. VRU activity and interventions have been co-developed with or informed by young people and communities and, as a result, have led to the intended outcomes.

The assessment, which was primarily informed through the triangulation of interviews with multiple stakeholders, covers progress since year ending March 2020 (the first year of VRU funding). It is followed by a discussion of the key enablers and barriers to young people and community engagement, which may help to explain the different levels of progress across the VRUs.

5.1 Progress towards effective engagement with young people and communities

Overall, there was evidence of some progress towards effective engagement with young people and communities. All VRUs had undertaken some engagement work with young people and communities, which had been effective to varying extents across the VRUs. The type and intensity of engagement with young people and communities undertaken by VRUs varied. All groups of stakeholders have identified engagement as an area that requires further development to ensure it is used effectively to influence the VRUs’ response to violence, and that this meets the needs of young people and communities.

In year ending March 2020, VRUs facilitated most engagement activities through existing platforms. Whilst VRUs made further progress in year ending March 2021, the pandemic significantly impacted engagement with young people and communities. Since then, all VRUs have continued to strengthen their engagement approaches, with most substantially increasing engagement activities and, in many cases, developing more formal structures to embed engagement processes within the VRU. There is some evidence that engagement has improved young people’s and communities’ perceptions of VRUs, and that young people and communities have been involved in the co-production of interventions. However, evidence of the effectiveness of this work is limited and several barriers made it challenging for the VRUs to progress as much as they had aimed for in this element of the whole-systems approach.

The following sections discuss the progress against each stage of engagement with young people and communities.

5.1.1 Engagement with young people and communities informs the VRUs’ response to violence

The Home Office expected all VRUs to engage with relevant young people and communities to inform their response to violence, and through a range of different approaches. These include through established groups, more informal ad hoc activities or with young people involved in interventions to gain their views in order to integrate these into VRU operations and delivery. This engagement aimed to ensure that the relevant audiences have a well-developed understanding of the VRU’s aims and, as a result, could provide their views of the prevailing issues and their causes, and in turn inform the VRU’s approach to addressing SV. Engagement with young people and communities was an integral part of the whole-systems approach in terms of the VRU being able to understand the issues locally through their perspectives and therefore embed practices tailored to the needs of the community.

Overall, all VRUs had made some progress to strengthen how they engaged with relevant young people and communities, which informed VRUs’ response to violence. For example, the evidence showed that most VRUs had aimed their engagement activities at specific geographical areas identified through the SNAs as being hot spot areas or those where specific at-risk groups were likely to be located. However, the effectiveness of this engagement varied among VRUs, which can largely be explained by the level and type of engagement delivered. The types of engagement VRUs delivered included established research or working advisory groups of young people and community representatives; local organised events to consult with young people and the community; through social media channels; through feedback surveys; and more targeted feedback sessions, where young people were involved in an intervention or initiative. Some VRUs used a mixture of these approaches to engage with young people and the community, whilst some engaged less, using only one or 2 approaches.

In VRUs where more limited engagement had taken place, none had a consistent or systematic approach to engaging young people and communities. In these cases, there was often a lack of capacity to drive the engagement forward and no strategic approach in place to monitor engagement. Whereas VRUs with more established and comprehensive engagement practices in place, including having a core strand lead responsible for the engagement activities, and those that had adopted a multi-faceted approach, generally had made more progress; for example, establishing young people or community groups which met monthly to discuss local challenges. Additionally, for a few of the VRUs, recruitment of community and voluntary sector representatives onto strategic and operational boards had led to more locally informed decision-making.

Evidence drawn from many of the strategic and operational stakeholders in the VRUs that had made more progress showed that their engagement activities had improved their understanding of the needs of young people and communities. With the operational stakeholders, this appeared to be specifically linked to the use of young people and community advisory groups, which were seen as beneficial in building an understanding of local concerns and priorities, and ways to collaborate on tackling these challenges.

“So really, the Peer Action Collective has provided us with a resource which is the perfect setup for us to be able to better understand what the causes of the children and young people who are violent and to really give them an opportunity to produce and design the solutions that they think are going to work in the social action arm of that project.”
Operational stakeholder

In addition, some VRUs had effectively used the results of their engagement activities to inform their violence strategy and approach, including:

• co-production of a renewed violence RS with young people to ensure it was targeted towards those in need
• young people and communities informing the SNA
• findings from engagement activities were used to shape discussions with key stakeholders to influence strategic thinking, such as in core VRU meetings and with the Police and Crime Commissioner (PCC) and police during strategic meetings; VRUs often considered topics discussed during engagement group sessions (such as exclusions, pupil referral units, home life, unemployment, young women and girls being drawn into the violence) in their strategic approach, and highlighted as concerns of a small proportion of young people and community representatives in the area

“Yes I think that the [VRU’s] work is very valuable, and they’re always liaising with us I think on a frequent basis, I know every couple of months we’ve spoken to them… and they’re very open to listening to that and then developing on certain strategies and planning. So I think, they’ve always been open to discussions and helpers and make some very impactful changes and work so I think, once again, they do very valuable work and it’s been very helpful.”
Young person involved in youth group

• involvement in decision-making relating to the commissioning of specific interventions or investment decisions; this involved young people influencing the VRU funding opportunities; for example, young people were involved in the interviews with prospective delivery providers, the discussion meetings, and shaping the priorities of the intervention

“They’ve got 40 young people that they are working with that have all got lived experience around serious violence and knife crime and the VRU is working directly with that [VRU16-Area6] youth group to engage with local people and help model what interventions we need to put in place.”
Strategic stakeholder

• the co-development of a Youth Violence Prevention Strategy; co-development involved young people who were already engaged in partners’ services to understand the main causes of violence and potential solutions
• young people shaping VRU social media responses to engage with the wider community about SV in the hotspot areas

5.1.2 Young people’s and communities’ perceptions of VRU/partners and interventions

One aim of the engagement activities outlined above was to improve the perceptions of young people and communities towards the VRU and its partners and interventions. Indicators of this would include whether young people and communities understood the aims of the VRU, whether young people and communities felt listened to, and if the engagement had increased awareness and trust in the VRU and partners.

Overall, progress was mixed across the VRUs, where all but a few had made only some or limited progress towards improving young people’s and communities’ perceptions of VRUs, partners and interventions. This was supported in the evidence gathered across relevant stakeholder groups, who often reported that further work was required to increase levels of awareness and an understanding of the purpose of VRUs.

For the few VRUs where there was some evidence that engagement with young people and communities had improved their perceptions of VRUs and partners, this had consistently happened because of VRUs working closely with an established group of young people or community representatives, such as a Youth Commission or Youth Advisory Group. Based on the small number of interviews conducted with young people and community representatives, there was evidence that young people involved in these groups felt listened to and that their views had contributed to the VRU’s understanding of violence and its drivers.

“I think, for me, the conversations I’ve had, they do seem to be in it for the right reasons. I’ve been involved with some things in the past where it’s like, ‘Is it tokenistic? Are they just speaking to us to tick a box?’, or whatever, but it does seem like they genuinely do care and they want to do the right thing and reduce violence. So, yes, I would say that I trust them. At least, the people that I’ve spoken to in the organisation.”
Young person involved in youth group

Conversely, young people involved in interventions delivered through community organisations generally had less awareness of the VRU. However, stakeholders did not consider this to be a problem because young people had often built positive relationships with delivery partners. Additionally, community organisations often preferred to disassociate themselves from the VRU to increase engagement with those identified as most at risk; there were concerns that if those at risk perceived the VRU as an extension of the police or statutory services, it may discourage them from taking part. Most young people involved in interventions reported that their feedback and opinions during interviews were considered and they felt listened to by delivery staff.

“We did a team away-day in September and one of the tension points that came to the surface just among the core team was around to what extent do we want people to be aware of the VRU and to what extent actually, is it okay for the VRU as a bit of a ghost entity from the point of view of communities? The key thing is that actually we know that their perception around, or their fear of crime and violence is reducing.”
Strategic stakeholder

There was some evidence from young people involved in some of the advisory groups and from frontline workers that engagement had improved the relationships between the police and young people in the area.

5.1.3 VRU activities co-developed with and informed by young people and communities

The Home Office expected all VRUs to co-develop some of their activities and interventions with young people and communities to ensure they meet the specific priorities and needs of each area.

Despite a strong commitment across VRUs to engage and involve young people and communities in co-producing their activities, progress has been extremely varied. Progress was most evident within the VRUs that had co-produced some of their intervention-related activity with young people and community representatives.

Although the level of co-production varied considerably, it was most often used to influence the design of a new intervention, or adapt and improve existing interventions, or to gain feedback from young people and the community on how to target hotspot areas. However, there was no evidence of VRUs using a systematic approach to co-production activities and instead they were more ad hoc and influenced by whether it was appropriate to modify interventions. Co-production activities were mostly led by either the VRU or delivery partner, with most young people and/or community representatives brought into the design stage at specific points to influence the intervention or initiative.

Several frontline workers reported that the co-production of interventions had led to activities being more effective, and had developed an understanding of the local communities to influence the design of interventions based on the local context. However, the evidence to support this remains limited because the interventions were still in the early stages of delivery.

“Normally, you see bits of funding and you just go for it, and you try and tailor what you do to suit the funding need at that time, whereas this one said, ‘Right, we’re not going to fund anything yet until we’ve listened to quite a lot of the community, until they’ve told us, what are their needs?’ [The community’s views informed how the programme of work is designed.] In terms of the design aspect, I feel like that’s different.”
Frontline worker

5.2 VRU approaches to young people and community engagement

As set out above, one of the key enablers of effective young people and community engagement centred upon the recruitment of dedicated resource to lead this strand of work. This tended to involve the appointment of an established community leader/representative into the core VRU team, who often had lived experience of SV and could therefore rapidly build credibility to influence the VRU’s activities, and who was associated with increased legitimacy within local communities.

“I think if [the communities lead] wasn’t in the VRU it might not have been as successful, because he feels like he’s from here, he feels like one of us. The relationship is easy to get with [name], because he’s been a youth worker for so long. He’s got shared concerns and shared interests. He gets it, he gets it. Not many people could have got [the communities lead] role, and we would have trusted him. We don’t trust easily, around here… over years we’ve learnt not to really trust authority, but they came in. I forget that the VRU is so statutory, because of the relationship we have with him.”
Community organisation leader

It was also evident that those VRUs which developed new or sought to enhance existing young people advisory groups and networks had made more progress, as these groups had been key in building awareness and trust amongst their peers and local communities. The VRU had recruited many of these groups directly, rather than by the police or the Office of the Police and Crime Commissioner (OPCC), which was also reported to have been a factor contributing to their success.

The involvement of young people in the design of interventions in partnership with delivery partners, rather than the VRU, was also seen as beneficial due to the community organisations having more established relationships with local residents. Several frontline workers reported that VRUs had encouraged delivery partners to involve young people in the co-design of interventions and to capture their feedback throughout the intervention. Stakeholders thought there were benefits to engaging with young people and communities in the co-production of interventions to help tailor them to the needs of the local area.

“So we commissioned a voluntary sector organisation that’s very well-respected locally. They basically pulled together a design group that comprised of mixed ages but people that were working or volunteering in grass root community groups in the 3 priority areas that this intervention is for. So, they were drawn from those communities. Then those people involved in the design process did the engagement with young people. So, that was the way that particular co-design process worked. Then we commissioned it.”
Operational stakeholder

Conversely, there were several barriers which VRUs faced in achieving effective engagement with young people and communities, some of which were similar to challenges that had arisen in year 2 (April 2020 to March 2021). This included the pandemic causing significant delays and disruptions to the delivery of face-to-face engagement activities, and the short-term funding cycle inhibiting the sustainability of meaningful engagement with young people and communities, as the annual cycle rarely allowed time to embed engagement approaches or establish cyclical feedback practices.

VRUs also often reported they had found it challenging to engage young people considered ‘hard to reach’, especially those who were most at risk of living in entrenched cycles of violence, and to determine who and how many individuals to engage to bring about meaningful engagement.

“We know that our cohort comes from deprived areas with difficult families and actually they’re not the people that really want to engage with us. So I guess, for me, it’s about, ‘How do we get that real, meaningful engagement with the people who we want to hear from, not the people who want to engage with us?’”
Strategic stakeholder

Case study: Effective community and young people engagement guiding needs-based commissioning

Across years 2 and 3, VRU A delivered its flagship intervention, a community-led pilot programme, which it documented it within its Response Strategy. The programme aimed to embed community and young people engagement/co-design, explicitly recognising their expertise and knowledge, to inform local needs-based commissioning. This included:

• identification of pilot areas based on analysis of 5 years of hotspot data of both victims and perpetrators of SV
• strategic engagement with local partners and asset mapping of the local area to develop a comprehensive understanding of the community context (for example, the local composition of housing and education)
• community and young people engagement, primarily conducted through focus groups, based on identifying community strengths, challenges and requirements; this learning was used to co-design the funding opportunity specifications to reflect local need and ensure buy-in
• the VRU communicated the funding opportunity to local community organisations; bidding criteria required them to work in ‘alliances’ to encourage partnership working and upskilling, rather than competition over funding
• a panel with community and young people representation assessed the applications to select the alliance to be awarded the bid
• before delivery began, the successful alliance met with the VRU and local evaluator to develop their delivery, monitoring and evaluation plans; this ensured effective mobilisation of delivery partners able to deliver and contribute to the evidence base

The VRU’s communities lead was identified as key to the success of the programme due to their experience within the voluntary, community and social enterprise (VCSE) sector, establishing trust within the community and challenging any negative perceptions about the programme.

There is some evidence to indicate that insight gained from this engagement had a wide reach, with an example provided of the community identifying eroding trust towards the police in a particular area (due to low police presence), and this being listened to and addressed, and the actions taken communicated via social media back to the public.

5.3 VRU contribution to engagement with young people and communities

Overall, the VRUs had facilitated some good progress in terms of the engagement of young people and communities. Whilst community-based organisations may have achieved some of this progress without the VRU, it is likely this would have taken longer. For several VRUs that have been a key facilitator in bringing voices together from the community and delivery organisations to inform their approach to violence, the contribution was more substantial and would likely not have occurred without the VRU.

However, stakeholders also recognised that further progress was required to improve young people’s and communities’ perceptions of VRUs, and enable VRU activities to be co-developed with them to achieve the intended outcomes. For example, there was limited evidence to support the key performance indicators around improved communities’ feelings of safety, and increased trust in the VRUs and partners.

It is anticipated that VRUs can build on the progress made with young people and communities in year 4 (April 2022 to March 2023), which will be aided by the 3-year funding cycle. It is also expected that the pandemic will have less of an impact on engagement activities.

6. Interventions

This chapter provides an assessment of VRUs’ progress and contribution towards commissioning and delivering interventions. Aligned to the programme-level ToC, this chapter considers progress against the following (connected/sequential) stages:

1. VRU interventions are strategically commissioned and evidence-based by drawing upon:

   • an accurate/comprehensive understanding of the drivers of SV and the at-risk cohort
   • existing evidence of the effectiveness of different interventions and/or contributing to the evidence through the monitoring/evaluation of interventions
2. This results in a response to violence where interventions are appropriately targeted and effective at supporting those most in need.
3. Ultimately leading to improved intended outcomes for young people and communities.

The assessment, which was primarily informed through the triangulation of interviews with multiple stakeholders and analysis of available interventions data, covers progress since year ending March 2020. It is followed by discussion of the approaches to effective commissioning and delivery of interventions.

6.1 Progress towards commissioning and delivering interventions

Overall progress towards both strategic and evidence-based commissioning has been good. In year ending March 2022, VRUs’ approaches to the commissioning and governance of interventions matured further and became more purposeful.

At the outset of the programme (year ending March 2020), VRUs commissioned their suite of interventions prior to / alongside the development of the SNAs and associated RSs. As such, VRUs had limited opportunities in year 1 to embed a strategic approach into their processes. Commissioning, therefore, varied across VRUs and was largely directed towards extending and enhancing existing interventions. Evidence from year ending March 2021 suggested commissioning had evolved in year 2 (April 2020 to March 2021) to become more strategic, and to some extent more evidence-based, across VRUs. This included more consideration given to why, what and where to commission interventions at the strategic level, and how to commission and govern these interventions at operational levels.

Particular areas of progress in year ending March 2022 included almost all VRUs engaging with partners to commission interventions based on identified needs, and subsequently more effective targeting of these interventions. Progress was more limited and mixed in relation to VRUs contributing to the evidence base (relying principally on descriptions of process rather than quantitative controlled impact evaluations), commissioning interventions informed by evidence of effectiveness, and maturing towards enabling a feedback loop from monitoring and evaluation to guide commissioning. This reflected some disparities in monitoring and evaluation capacity between VRUs.

The following sections discuss progress against each stage of commissioning and delivering interventions.

6.1.1 Strategically commissioning evidence-based interventions

Overall, good levels of progress in strategically commissioning evidence-based interventions were made in all but a few VRUs. This was broadly supported by evidence across a review of funding applications, SNAs, RSs and the qualitative interviews with relevant stakeholder groups.

Key indications of the effectiveness of VRUs’ commissioning functions were suggested by evidence of VRU capability to:

• bid successfully for additional funding, including from the Serious Violence Youth Interventions Programmes (early interventions), and lever in matched funding as an enabler to greater financial sustainability (for example, from a charitable institution, Core Member, or other funder to deliver interventions); however, significant variation was evident in the quality and depth of the bids received by the Home Office
• engage and draw upon the findings within the Youth Endowment Fund (YEF) toolkit^{[footnote 9]} to inform commissioning based on evidence of ‘what works’, although this could be strengthened
• appoint an external evaluator (for example, external partners and academic institutions) in a timely manner and contract for a robust evaluation programme, with some VRUs being delayed and significant variation evident in the evaluations commissioned; further reliance is needed on the toolkits and guidance published by the What Works Centres, especially the Early Intervention Foundation, the YEF, and the College of Policing ‘What Works Centre for Crime Reduction’

In year ending March 2022, VRUs engaged with partners to commission interventions. As each VRU was given the autonomy to determine how to allocate and prioritise how they spent their allocation, there was considerable variation in intervention spend across VRUs (see Table A.1 in Annex A). Following the trends from year ending March 2021, governance processes for commissioning tended to be proportionate to the scale of the funding allocated to interventions. Also, some VRUs devolved funding to the local level (these were often smaller pots of funding, such as to CSPs and community umbrella organisations).

Besides interventions that were commissioned using the ‘core’ VRU funding settlement from the Home Office, VRUs were provided with the opportunity to bid for additional funding from the Serious Violence Youth Interventions Programme Fund in year ending March 2022. The fund supported the delivery of 3 types of interventions: High-Intensity Therapeutic interventions; Teachable Moments interventions; and trauma-informed training. VRUs could submit up to 3 bids for grant funding across the 3 pots of funding. VRUs commissioned the following:

• 4 High-Intensity Therapeutic interventions, all of which were Cognitive Behavioural Therapy (CBT) interventions
• 19 Teachable Moments, with 8 providing interventions in police custody settings, 9 in hospital and A&E settings^{[footnote 10]} and 2 in educational settings
• for the 7 trauma-informed training interventions that were successful in their application, the relevant VRUs were aiming to reach thousands of professionals working with children and young people across different sectors, such as police, health, early years and education

The Home Office also worked with the YEF in year ending March 2022 to develop a co-funding option for 4 projects, which allowed for multi-year funding and enhanced impact evaluation.

The Home Office application guidance^{[footnote 11]} acted as a lever to embed a strategic and evidence-based approach to commissioning interventions within VRUs and, in relation to this, set out that the Home Office expected all VRUs to:

1. Monitor and report expenditure and progress quarterly to the Home Office.
2. Have a ToC in place, agreed by the Core Membership, which should drive selection and monitoring of local success measures.
3. Complete a self assessment in relation to the strength of evidence underpinning the VRU’s proposed portfolio of interventions as part of its Delivery Plan. This was to include whether the intervention appeared in the YEF Toolkit, any further evidence used, and any planned evaluation assessing the quality of methodology used for impact evaluation against the Maryland Scale (for example, level 4).

For the Serious Violence Youth Interventions Programme, the Home Office issued additional guidance^{[footnote 12]}, which also stated the expectation that VRUs would evaluate the interventions and adhere to more frequent monthly performance reporting on expenditure and progress.

In year ending March 2022, VRUs made good progress towards both strategic and evidence-based commissioning of interventions. Although extending and enhancing existing services is still evident in practice, the approach towards commissioning has matured further and become more purposeful. Almost all VRUs have engaged with partners to commission interventions based more on identified needs. This included drawing upon an accurate/comprehensive understanding of the drivers of SV and the at-risk cohort. The SNAs and RS provided an initial evidence-based foundation which set the focus and direction of travel for the VRUs.

“The starting point is the Strategic Needs Assessment… The data from the SNA is used to design projects - as the data provided informs communities an intervention would work best in. The data is the best force for knowing what to fund.”
Operational stakeholder

Particular progress in year ending March 2022 included several VRUs drawing upon more agile intelligence and data on trends in violence and the at-risk cohorts, the hotspots (and hot times) where incidents were most likely to take place to inform and target their portfolio of interventions, alongside evidence and tacit knowledge of what has worked well and the uptake of services. This was particularly important given the variation in the quality and static nature of the SNAs and RS (in particular, their ability to identify and inform response on hotpots).

Most VRUs also used, and weighed up, a range of existing evidence of the effectiveness of different interventions to inform decisions on commissioning, although this could be substantially strengthened. This included:

• actively considering the findings set out in the YEF’s Toolkit to feed into portfolio analysis and decisions, and the commissioning of interventions assessed as ‘high impact’ by the Toolkit
• wider published (and peer reviewed) evidence, guidance and toolkits from the What Works Centres, especially the Early Intervention Youth Fund, YEF, the College of Policing ‘What Works Centre for Crime Reduction’, and the evidence base on the Scottish VRU, influencing the adoption of some of its flagship interventions, such as Mentors in Violence Prevention programme and A&E Navigators across several VRUs
• local evaluation evidence, which was available for almost all VRUs; however, the quality of the evidence base generated by local evaluations was limited and variable, with some evaluations conducted internally, potentially without the methodology/limitations being explicitly presented; the evaluations undertaken had mostly been process evaluations, with some outcome and impact evaluations available
• tacit knowledge from Core Members and delivery partners on what has worked well and the uptake of services; one VRU gave an example of undertaking site visits to see how interventions were delivered and to discuss with local partners how to transfer best practice to their own setting

A few VRUs drew on outputs from consultations with community partners^{[footnote 13]} to influence the type and location of commissioned activities.

6.1.2 VRUs contributing to the evidence through monitoring/evaluation of interventions

Progress was more limited and mixed in relation to VRUs contributing to the evidence base (relying principally on descriptions of process rather than quantitative controlled impact evaluations) and enabling a feedback loop from monitoring and evaluation to guide commissioning. This reflected some disparities in monitoring and evaluation capabilities, capacities and practices between VRUs.

Most VRUs reported achieving progress in relation to monitoring and reporting, including improvement in the quality and timeliness of Home Office monitoring returns. This suggests a strengthening in wider management capabilities (including financial and project management). However, some stakeholders (in particular frontline stakeholders) identified that monitoring administrative procedures remained complicated and time consuming proportionate to the funding allocated. In several VRUs, there was still an identified need to ensure monitoring information from VRU interventions was of sufficient quality for use at VRU-level or for future use in evaluation (or for a plan to be put in place to improve monitoring quality). This includes the need for VRUs to create common outcomes frameworks underpinned by their ToC to guide all interventions towards achieving the same, broader aims, and enabling a feedback loop to proportionate outcomes-based commissioning.

VRUs also continued to commission local evaluations, which have generally been process orientated and/or pan-VRU strategic evaluations, although there are exceptions and some impact evaluations have been commissioned. Those VRUs with in-house evaluation capacity (for example, an evaluation lead) tended to be more mature and to have produced and developed an outcomes framework, and tended to be commissioning more comprehensive evaluations. In turn, this could inform outcomes-based commissioning.

In several VRUs, the qualitative interviews suggested there was a perception that some local evaluations commissioned had become better targeted at meeting evidence gaps over time. Some stakeholders described the process of VRUs actively understanding their evidence base, and then dedicating resource to add value at a local/national level, to capture evidence of the improving sophistication of VRU delivery. In particular, stakeholders identified the need to improve understanding of the impact of less well-evidenced activity (for example, one VRU supported pilot schemes and upscaled the rollout of successful pilots). As one stakeholder noted:

“We’ve really tried to keep an eye on what works and an eye on those types of interventions that have got a real clear evidence base. For those that are less tested, we have commissioned our academic evaluators to focus on them so that we’ve got a clearer picture… So, that’s going to come out in our academic evaluation and if the evidence suggests it doesn’t work, then we’ll stop doing it. If it does suggest it works, then we can share that learning.”
Operational stakeholder

As the above sets out, several VRUs were acting as generators of evidence which could allow for shared learning on the most effective approaches to preventing serious youth violence, where gaps were identified in the evidence base, which appropriate research bodies might use. For instance, the YEF ‘Evidence and Gap Map report’ (White et al., 2021) highlighted a paucity of evidence regarding contextual safeguarding, interventions which aim to prevent child criminal exploitation, child-focused criminal justice methods, and approaches that look to change systems that interact with children.

6.1.3 Appropriate targeting of VRU interventions

In all but a few VRUs, good progress has been made towards interventions being more appropriately targeted to support those most in need. This was broadly supported by evidence across funding applications (specifically, delivery plans, which varied in comprehensiveness and quality), monitoring information, and evidence from the qualitative interviews which captured the perceptions of relevant stakeholder groups.

Almost all VRUs were effectively engaging the intended target groups for interventions, as illustrated by the figures for actual delivery in Table 6.1 below. In relation to delivery models, the VRUs commissioned a portfolio of at least 200 interventions in year ending March 2022. They delivered a range of interventions, over a third of which focused on early intervention (40%, 79). Approximately one in 5 focused on either desistence (21%, 42), prevention (19%, 38) and therapeutic interventions (17%, 34)^{[footnote 14]}. This represents an increase in the overall proportion of early intervention delivery models delivered compared to year ending March 2021^{[footnote 15]} (particularly targeted at level 2 – potentially high-risk groups)^{[footnote 16]}, and a reduction in prevention models^{[footnote 17]}, suggesting greater targeting of interventions at those at increased risk. See Table A.2 in Annex A.

Table 6.1: Number of interventions delivered by target group and delivery model, all VRUs, year ending March 2022

Target groups	Prevention	Early intervention	Therapeutic	Desistance	Unknown	Total
Level 4: Involved in crime/violence	1	4	7	26	0	38
Level 3: Known risk/Known to services	3	20	13	11	1	48
Level 2: Potentially high risk	6	32	7	3	0	48
Level 1: Universal	28	26	7	2	0	63
Other	0	0	0	0	6	6
Total	38	82	34	42	7	203

Notes:

1. Please treat the data displayed in this table as indicative, as it is based on varying data quality and levels of completeness.

Besides the monitoring data, and despite limited local evaluation evidence on outcomes, there was emerging evidence from the qualitative interviews that stakeholders’ perceived interventions were becoming more effective.

This was through enhancing existing services, adjusting delivery when needed (for example, to changes in context due to COVID-19), and through being able to trial new approaches. Progress identified through the qualitative stakeholder interviews included improved identification and engagement of target groups; improved referral of young people to interventions and to the appropriate intensity of interventions; increased engagement among young people; and improved attitudes towards statutory services or willingness to communicate and engage with services or providers. One stakeholder noted:

“They’ll look at education data, they’ll look at whether siblings are known to services, they’ll look at the social care data, and they will do a community resolution assessment or prevention assessment and then an intervention plan. It’s not time-limited, we can have a young person get 2 sessions with a prevention officer, we can work with a young person intensively for 5 months. We won’t give a diluted service; we give the service that’s required for sustainable change.”
Frontline stakeholder

6.1.4 Young people and community outcomes of VRU interventions

Overall, there was emerging evidence that VRUs made some progress towards realising the intended outcomes of the commissioned interventions. At the time of reporting, an analysis of the monitoring information available indicated that interventions had reached 217,491 individuals across the VRUs^{[footnote 18]} (see Table 6.2). Direct comparison between year ending March 2022 and year ending March 2021 data is not possible due to differences in the data collection methods and incomplete data. However, Table 6.2 suggests that whilst VRUs reached fewer individuals in year 3 (217,491 in year ending March 2022 compared to 262,210 in year ending March 2021), this change has been driven by a significant increase in targeting towards cohorts at risk level 2 (potentially high risk) and level 3 (known risk/to services), and a reduction in the level 1 (universal) target group being supported (see Table A.2 in Annex A). This implies that the commissioning of interventions has become more targeted over time, likely due to VRUs drawing upon greater insight of their at-risk cohorts.

The data presented are based on varying quality and levels of completeness and includes an analysis of outputs (number of people supported) rather than outcomes. However, they do indicate some high-level insight into VRU commissioning and variance in intervention average spend across target risk groups.

Firstly, interventions focused on level 1 (universal) target groups account for the highest proportion of projects commissioned in year ending March 2022 (63, 31%), reaching 51,290 people, with an average spend per person reached of £89. Whereas interventions focused on level 4 (involved in crime/violence) target groups account for the lowest proportion of projects commissioned (38, 19%), with 14,979 people reached, and an average spend per person reached of £232. This indicates that the higher intensity of delivery, and therefore cost of interventions, aligns with the highest risk cohort compared to the lowest risk cohort.

Secondly, there have been the same number of commissioned interventions targeted at level 2 (potentially high risk) and level 3 (known risk/to services) risk cohorts in year ending March 2022. However, level 2 reached a higher number of individuals (81,416) compared to level 3 (66,784). The cost ratings from the YEF Toolkit (which gives a general indication of the cost of approaches (YEF, 2022)) would suggest that interventions targeted at level 3 (known risk/to services) and level 2 (potentially high risk) may be expected to have higher spend per person. For instance, in the YEF Toolkit, A and E navigators, interventions to prevent school exclusions and social skills training interventions have an estimated average cost per participant of £500 to £1500. CBT and mentoring have an estimated average cost per participant of £1500 plus^{[footnote 19]}. Yet the average spend per person reached for interventions targeted at level 3 risk cohorts (those with known risk to services) is lower at £71 per person reached, compared to £81 for interventions targeted at level 2 risk cohorts (those that are classified as potentially high risk). This may result from COVID-19 related disruptions that continued to be experienced in year 3 (April 2021 to March 2022) and should be monitored as the programme continues to mature to ensure investment is targeted appropriately.

The analysis also demonstrates the importance of interventions delivering at the expected level of intensity, and recording this accurately, so that costs can be accurately determined. VRUs recorded most of the data in Table 6.2 as part of their monitoring returns to the Home Office. But one VRU provided an estimated distribution of engagement across different target groups, which had to be used to establish spend per person. The VRU indicated it had reached substantially more individuals at level 4 than the other VRUs combined. Including the estimates meant that the average spend per person reached for those in target group level 4 was £232, whereas it averaged £590 across the other 17 VRUs. Spend per person for other target groups also fell with this area’s estimates, although not substantially – level 3 spend per person was £73 across 17 VRU areas and £71 across the 18; level 2 was £106 across 17 areas and £81 across 18; and level 1 was £95 across 17 areas and £89 across all 18. There is no indication these figures are wrong, but they reinforce the need for scrutiny to ensure best practice is being delivered.

Table 6.2: Number of projects delivered, total spend to date (January to April 2022) and number of individuals supported, by target group, year ending March 2022

Target groups	Spend to date Jan-Apr 2002	No. of projects	No. of people supported [footnote 20]	Average spend per project	Average spend per person reached
Level 4: Involved in crime/violence	£3,470,181.73	38	14,979	£91,321	£232
Level 3: Known risk/Known to services	£4,764,181.07	48	66,784	£99,254	£71
Level 2: Potentially high risk	£6,627,144.62	48	81,416	£138,066	£81
Level 1: Universal	£4,575,266.80	63	51,290	£72,623	£89
Other	£807,440	6	2,673	£134,573	£302
Total	£20,244,214.22	200	217,491	-	-

Notes:

1. Please treat the data displayed in this table as indicative, as it is based on varying data quality and levels of completeness.

Although it is not possible to comment on the effectiveness of the individual interventions delivered at this stage^{[footnote 21][footnote 22]}, in almost all VRUs the young people interviewed all spoke positively about their engagement in interventions and frontline stakeholders reported they felt they were working well with young people to meet their needs and build trust. This was particularly the case for Serious Violence Youth Interventions, which were delivered at a higher level of intensity. A theme throughout stakeholder and young persons interviews was the importance of the professional working with the young person being relatable (for example, a few people described this as “cultural competency” or “cultural sensitivity”), and being reliable (“doing what they said they would do”) in building up a trusting and positive relationship.

Given the nature of the interviews conducted with young people^{[footnote 23]}, it is not possible to draw firm conclusions on outcomes achieved by different types of interventions, which local evaluations in year ending March 2022 should be able to indicate.

Evidence captured through the qualitative interviews of VRUs’ progress towards achieving outcomes related to reduced offending and improved community safety was limited. However, there was some evidence of progress in relation to the achievement of longer-term outcomes set out in the programme’s ToC around the improvement of life outcomes for young people for almost all VRUs based on the small number of interviews with young people and the stakeholder interviews. This included outcomes that related to:

• improved confidence and self-esteem:

“It’s helped me with my confidence, speaking out in front of a group.”
Young person

• improved motivation/positivity/resilience and overall wellbeing, including mental health:

“It’s helping me with my mental health… I know that when my mental health’s deteriorating, I’ve got to do something about it. And obviously, going to these meetings and speaking, even just speaking for 10 minutes with him, he’s just helping me out. Given me a bit of a booster, do you know what I mean?”
Young person

• improved relationships with family, friends and others; and movement towards the reduction of young people not in employment, education or training (NEET):

“What we can see in terms of outcomes is that a significant number, over 50%, have either completely stopped their drug use or reduced it to such a level that it’s really occasional. So, actually, school engagement has improved, or access into mental and emotional health support has improved, or relationship with social worker has changed, or conflict in family has improved.”
Frontline worker

• improved behaviours (for example, anger management):

“I stopped having fights with people as much.”
Young person

• to a lesser extent, (apparent) reduced likelihood to re-offend:

“The people I was involved with I completely stopped talking to them. I’ve realised by [VRU worker] speaking to me, I could’ve been getting into stuff where I wouldn’t want to be. My behaviour has changed and mentally I have changed. I have stopped going off the rails ignoring what my parents say. Because there is no point. I used to lie all the time, now there’s just no point lying. Mentally I have less stress on me. I went to the hospital the other day, because of all the stuff that’s been building up, I got really anxious. But when I speak to [VRU worker], it doesn’t hurt no more, it’s like a weight off my chest.”
Young person

The vignette below provides a case study based on our interview with one young person.

Case study: A&E Navigators intervention (based on an interview with one young person)

Eleanor (name changed) took part in the Navigator project, which places youth workers in the A&E departments of 4 hospitals to support young victims of violent assault (including knife-related crime). The youth workers support the young person for a period of approximately 6 weeks (longer if needed), mentoring them, identifying their needs and connecting with local organisations that can best support them long term.

Eleanor was referred to the Navigators programme during an A&E appointment for injuries sustained in a fight. Eleanor related that her Navigator made her feel comfortable in that first meeting, and that the relationship had built trust over time.

“It’s just like one of those people you can speak to and not feel judged on what you’re saying… someone who’s actually willing to be there for you, and someone to speak to when you need help.”
Young person

Eleanor expressed that the programme had helped to improve her confidence, self-esteem and mental health by encouraging her to reflect more positively on herself. She also mentioned that the programme had helped her with anger management and school attendance. Overall, Eleanor said she would recommend the programme to others.

“I’ve felt more confident in myself because [the Navigator] tells me not to put yourself down and that. And it’s made me feel like more of a person and more happier.”
Young person

“I’ve learnt how to control my anger and how to speak to people without lashing out.”
Young person

“When I’m in school and I don’t go in my lessons, I’ll be, ‘I skipped my lessons’. But [the Navigator] told me to start going in them, and then obviously I’ve listened to her, and I’ve started to be well-behaved in school.”
Young person

6.2 VRU approaches to commissioning and delivering interventions

VRUs adopted a variety of commissioning approaches. For instance, the model of working may have included an initial analysis of need, followed by interventions being commissioned through the VRU communicating a call for proposals with applications/competitive bids coming from community organisations, which were then assessed based on VRU criteria. There may also have been ‘check and balance’ processes to ensure that commissioning met procurement requirements and evidence of need supported decisions, whilst avoiding duplication of provision. However, VRUs ranged in the formality and maturity of these commissioning processes and, following the trend from year ending March 2021, governance processes for commissioning tended to be proportionate to the scale of the funding allocated to interventions. Also, a few frontline stakeholders fed back the importance of flexibility within commissioning processes to enable adaptation to local need and changing contexts.

In terms of the VRU model influencing the approach to commissioning adopted by the VRU (for example, both centralised and localised models of working), there was some evidence to suggest that potentially a more centralised approach enabled more formalised evaluation infrastructure and processes to develop. Several VRUs adopted a more decentralised approach to the commissioning of interventions, particularly where they covered heterogenous areas which served diverse communities. Several VRUs also adopted a scaled approach to commissioning, where they commissioned area-wide/large-scale interventions centrally, and smaller, more community-based interventions locally.

In analysing VRU approaches to commissioning and delivering interventions, both operationally and strategically, there were some emerging trends that highlighted key contextual factors which supported greater progress. Pre-existing VRU / multi-agency partnerships were more likely to be able to draw upon existing interventions or partnerships with organisations with a ‘proven’ delivery model in place. In particular, for Serious Violence Youth Interventions Programme Funding, the VRUs tended to commission interventions from existing services with established referral pathways. Also, in several cases, alignment with Grip supported effective and targeted commissioning. Strategic and operational VRU staff perceived this alignment positively; for example, focusing on hotspot areas, and targeting preventative work in locations where Grip operations also took place.

In terms of modifiable factors which supported greater progress (that is, factors which VRUs had more control over), sometimes VRUs had aimed to bolster evaluation capability by developing partnerships with research bodies with the necessary evaluation expertise; at least one VRU appointed an evidence and evaluation lead, and another involved their local evaluator in the design phase of interventions. There were further examples provided of VRUs aiming to build evaluation capability in the system through the use of training, workshops and developing tools. Likewise, a few VRUs had aimed to further formalise and mature their commissioning process and develop their evidence review and evaluation infrastructure. One VRU had created an ‘Evidence Hub’ to draw upon a wide range of evidence sources to inform their commissioning approach and yes/no-go decision-making. Another had developed an intervention assessment and prioritisation tool (matrix) for commissioning VRU interventions to embed a consistent approach. At least one VRU had a documented monitoring, evaluation and learning framework (see case study below).

Case study: Example of successful embedding of Monitoring, Evaluation and Learning Framework

VRU model: Centralised.

Context: VRU A was originally set up as 3-year programme, which built on an established Strategic Partnership Board set up prior to the VRU.

Core team: The core team is co-located and includes an evaluation lead recruited in year ending March 2021 for the next 3 years to ensure that learnings are shared and taken on board throughout the year.

VRU A has developed a Monitoring, Evaluation and Learning Framework which is documented within the RS, and which sets out the intentions for 6 key elements:

1. The evidence base.
2. Local datasets.
3. Performance monitoring.
4. Insights and feedback.
5. Internal and external evaluation.
6. Knowledge exchange.

The framework has enabled the following:

• the VRU used its overarching ToC to identify key outcome indicators relating to the whole-systems approach; it then developed a monitoring and evaluation framework to progress intervention data collection to a standard where the VRU can better understand outcomes at intervention- and VRU-level
• this outcomes framework underpins the design, commissioning and delivery of interventions and local evaluations
• in year ending March 2022, there was some evidence that dissemination of monitoring and evaluation evidence on short-term outcomes had enabled an effective feedback loop to utilise monitoring and evaluation findings to guide investment selection and delivery:

“Because we’ve got the monitoring frameworks in place, you’re forever scrutinising it, in a good way.”
Operational stakeholder

• clear guidelines and expectations for delivery partners, with the VRU building evaluation capability through resourcing training and workshops (which it plans to expand to wider community organisations)

“We’ve worked with the local Violence Reduction Information Network at the [University], they’re developing a digital training package which is very much what is evaluation, how can you do it and it’s tailored to crime and community safety. Again, we’re thinking about our broader CSP partners. When we now start to do our workshops around this is the data and the dashboard, a core part of that will be make sure you’re still monitoring it and evaluating it, and this is how you can do it.”
Operational stakeholder

Looking forward, it is important to also consider some factors that hindered progress. In terms of contextual factors, the short-term nature of the funding cycle was a key issue to sustain engagement of providers and participants, including the recruitment and retention of frontline workers to deliver interventions. This was particularly the case for Serious Violence Youth Interventions, with examples provided of difficulties recruiting therapists and trained psychologists due to the funding being expecting to end in March 2022. Likewise, the timing of Grant Agreements was identified as a key driver for some VRUs to be less strategic in commissioning than intended, needing to mobilise interventions and external evaluations at pace. Frontline stakeholders in particular described difficulty with mobilising at short notice. This was particularly acute for Serious Violence Youth Interventions given the minimum grant value^{[footnote 24]}.

“We were given the funding last minute and it had to be spent so we just went out to those organisations we knew that could deliver against it within the shortest period of time.”
Operational stakeholder

“So, you’ve got a year’s worth of money in a 6-month period which was frustrating.”
Serious Violence Youth Intervention delivery partner

Finally, COVID-19 still limited progress in year ending March 2022, reducing the effectiveness of some interventions. Indeed, the impact of responding to the pandemic on statutory service providers’ budgets was significant, leading to uncertainty over their capacity to deliver core services and refer effectively into interventions.

In terms of modifiable factors which hindered progress (that is, factors which VRUs had more control over), several VRUs had still not yet struck the appropriate balance in relation to oversight of their commissioning, which was felt to be problematic and requiring further consideration. At least one VRU acknowledged that although their CSPs were asked to locally evaluate interventions and had been offered monitoring and evaluation support from their VRU, take-up was poor and the quality of evaluation outputs was mixed. It may also be challenging to produce robust evaluation data if many small-scale interventions are funded with low-level effects.

For commissioning local evaluation, there were applicability challenges associated with implementing certain evaluation methodologies, in particular counterfactual impact evaluation. A combination of high-quality (external) evaluation of select interventions (for example, where counterfactual impact evaluation is feasible) and good monitoring / internal evaluation of other interventions may help ensure a proportionate evidence base is developed. Likewise, when planning for evaluation, there was a need to consider the alignment and timing of commissioning evaluations against delivering interventions. Commissioning of the YEF evaluations highlighted how long it took to get an evaluation off the ground and gives an indication of the level of spend needed for robust impact evaluation at the intervention level. Overall, their evaluation capability and knowledge constrained many VRUs to commission meaningful evaluations themselves, and they should draw on the expertise available through the What Works Centres.

6.3 VRU contribution to commissioning and delivering interventions

VRUs contributed centrally to evidence-based commissioning and delivery of interventions, including for Serious Violence Youth Interventions. In particular, the qualitative interviews suggest the perception that VRU leads had an important role to play in translating national requirements and championing evidence use and evaluation, and the Core Member partnership structure allowed VRUs to tap into a wide pool of delivery providers able to engage targeted audiences due to pre-existing relationships within the community. Overall, VRU contribution was identified through:

• progress towards commissioning a mixed portfolio of interventions, allowing for evidence-based and innovative approaches, whilst ensuring less well-evidenced activity, would be evaluated to improve understanding of its impacts
• harnessing the expertise of the VCS and supporting delivery providers who were well-trusted and embedded in the target communities; this included VRU activities such as translating national guidance and evidence to apply to local partners; a few stakeholders reported wanting more consistent communication and steer from their VRU; however, almost all stakeholders fed back that their relationship with their VRU remained highly positive; this was particularly the case for Serious Violence Youth Interventions:

“What has worked well has been having the opportunity, having a VRU who are genuinely really supportive with the intervention.”
Serious Violence Youth Intervention delivery partner

• encouraging knowledge sharing within (and to a lesser extent across) VRUs, including between intervention providers and with appropriate research bodies:

“It is about sharing good practice and the VRU is a really good platform and network to do that.”
Frontline worker

• further embedding monitoring and evaluation plans throughout commissioned interventions and appointing external evaluation providers
• in several VRUs, there was some evidence of a feedback loop whereby monitoring and evaluation of interventions had informed subsequent delivery or commissioning and, to a limited extent, had informed SNAs/RS; for several VRUs, however, monitoring and evaluation was still maturing, and further intervention selectivity and intervention fidelity checking was required; those VRUs were not in a position to implement feedback loops and proportionate outcomes-based commissioning

7. Conclusions and recommendations

This final chapter presents conclusions from the year ending March 2022 evaluation on the overall contribution of VRUs to a whole-systems approach to prevent violence. The chapter also provides recommendations for the VRU programme moving into year ending March 2022 to year ending March 2023 and beyond.

7.1 VRU contribution

In year ending March 2022, VRUs showed signs of maturing and becoming embedded in, and recognised as leading and coordinating, local responses to prevent violence. This supported increased engagement and commitment from partners (such as, local agencies/organisations) as the role and value of VRUs became clearer. Across VRUs, it was evident the scale and focus of multi-agency working to prevent violence observed would not have been achieved without the VRUs.

The VRUs’ approaches to sharing and analysis of data progressed. This further supported their multi-agency response by enabling VRUs’ and partners’ efforts to be better targeted at those in need and reduce inefficiencies.

Progress was also evident in the commissioning of evidence-based interventions and targeting these at those in need. This was supported through wider VRU structures and functions, and clearer requirements and guidance on commissioning and generating evidence of interventions. Progress was somewhat more limited around the monitoring and local evaluation of interventions.

Engagement with young people and communities was progressing but less so than other elements of the whole-systems approach. VRUs had sought to catch up from engagement time lost due to the COVID-19 pandemic and to engage VSOs.

In terms of reductions in violence, there were no statistically significant effects on the primary SV outcomes of hospital admissions for sharp object violence or homicides. However, the direction of homicides after a relatively short period, alongside reductions in secondary SV outcomes, was encouraging. Stakeholders were of the view that the whole-systems approach of VRUs was contributing to this, but it was difficult to determine direct attribution owing to the complex landscape in which VRUs operate.

Whilst overall progress towards the whole-systems approach, and the indications of reductions in violence, were encouraging, VRUs recognised there is still work to do. This included continuing to build on their strengths and addressing areas where progress has been more limited. The engagement of some partners, in the multi-agency response and/or in sharing data, remained a barrier for some VRUs.

7.2 Recommendations

Where feasible and appropriate (for example, within their specific context), key recommendations for VRUs include:

• continue to engage with a wide range of partners, in particular, the Core Members from health and VSO partners, where engagement may again have been limited, to further the whole-systems approach
• linked to the above, work with partners to ensure their role within the response to violence is strengths-based, proportionate and manageable; this would help ensure meaningful engagement and added value from partners, including those with more limited capacity; in other words, VRUs should continue to focus on the quality of partnerships, which might require focusing on key partners initially and gradually involving others
• as far as possible, focus on providing non-financial support to partners to support and develop their capacity, which will support more sustainable and longer-term impacts
• aim for access to at least sub-area level data from key sources so that interventions and wider activity (for example, multi-agency responses to hot spots) can be targeted at areas/groups most in need; VRUs could facilitate this through sharing learning and tools (for example, standardised data-sharing protocols)
• ensure there is sufficient expert capacity available to, or within, the VRU to negotiate access to data and provide actionable insights from the analysis; this could include expanding data teams, working with external experts and/or learning from other VRUs
• ensure there is representation in the Core Membership from community leaders and representatives of groups / grassroots organisations that can provide valuable insights to (and experience of working with) the at-risk cohort
• focus on co-development with young people and communities to ensure that evidence-based interventions and VRU activities are tailored to local contexts and needs; consider drawing on the evidence base to inform engagement approaches so they effectively engage different groups; where possible, consider a dedicated community engagement role to facilitate this
• ensure that updated Response Strategies clearly articulate how interventions meet identified needs (based on data), that the VRUs capture these in both their Theory of Change (ToC) and Evaluation Plan/Strategy, and that progress towards meeting these needs can be monitored; this will help ensure interventions are evidenced-based and developed to meet local needs
• ensure that intervention portfolios include interventions where there is strong evidence of effectiveness to maximise the impact of the VRU; this could include interventions detailed in the YEF Toolkit (or similar) and local evaluation findings; where existing evidence is more limited, VRUs should ensure they conduct appropriate monitoring/evaluation to build the evidence base
• develop monitoring and evaluation strategies; this would improve the consistency, comparability and quality of evidence generated, which would provide learning for future commissioning decisions; where possible and appropriate, VRUs could seek to strengthen their evaluation capacity through increased collaboration with the What Works Centres and research bodies (for example, local universities)

Recommendations for the Home Office, which would support the above, include:

• exploring with the Department for Health and Social Care whether they can encourage and support health partners in VRU areas to increase engagement and share data; this should include clarifying what data can and cannot be shared under current/upcoming (for example, SV Duty) legislation and clearly communicating this with VRUs and partner agencies
• where good practice has been identified (for example, trauma-informed training, standardised data-sharing protocols), support the sharing of tools and resources for wider implementation
• increase support and guidance to VRUs to become more sustainable (for example, types of support provided to partners, bid writing, and demonstrating value to partners)
• exploring if the Home Office can work with the NHS to provide (extracts of) national administrative data to VRUs directly (for example, from NHS Digital rather than through specific trusts) at a sub-area or individual level
• revisiting and increasing the policy focus on the expectations around community and young person engagement
• develop guidance and templates for effective VRU intervention monitoring and evaluation strategies, emphasising the importance of accurate data records

Annex A: Additional data tables

Table A.1: Number of projects, spend and individuals supported, by target group, year ending March 2022^{[footnote 25]}

VRU	Funding allocation	Intervention spend total	% of total funding
London	£7,000,000	£5,491,028	78%
Merseyside	£3,370,000	£2,112,824	63%
Greater Manchester	£3,370,000	£2,150,000	64%
West Midlands	£3,370,000	£1,731,056	51%
West Yorkshire	£3,370,000	£1,691,659	50%
Northumbria	£1,600,000	£1,000,564	63%
South Yorkshire	£1,600,000	£ 871,226	54%
Thames Valley	£1,160,000	£968,962	61%
Kent	£1,160,000	£519,047	45%
Avon & Somerset	£1,160,000	£686,037	43%
Essex	£1,160,000	£592,680	37%
Lancashire	£1,160,000	£392,876	25%
Sussex	£880,000	£500,705	57%
Hampshire	£880,000	£487,128	55%
South Wales	£880,000	£330,568	38%
Bedfordshire	£880,000	£281,049	32%
Leicestershire	£880,000	£266,360	30%
Nottinghamshire	£880,000	£430,493	49%

Table A.2: Number of interventions delivered by target group and delivery, model year ending March 2021^{[footnote 26]}

Target groups	Prevention	Early intervention	Therapeutic	Desistance	Unknown
Level 4: Involved in crime/violence	3	6	12	37	0
Level 3: Known risk/to services	14	16	19	11	0
Level 2: Potentially high risk	11	11	5	4	0
Level 1: Universal	41	17	6	0	0
Other	0	1	0	0	2
Total	69	51	42	52	2

Notes:

1. Please treat the data displayed in this table as indicative, as it is based on varying quality and levels of completeness.

Table A.3: Number of individuals supported by VRUs in year ending March 2021 and year ending March 2022^{[footnote 27]}

Target group	No. people reached year ending March 2021	No. people reached year ending March 2022
Level 4: Involved in crime/violence	6,412	14,979
Level 3: Known risk/to services	18,847	66,784
Level 2: Potentially high risk	18,698	81,416
Level 1: Universal	214,561	51,290
Other	3,692	2,673
Total	262,210	217,491

Annex B: Technical annex

This technical annex provides detail on the data sources and analytical approach for the quasi-experimental designs implemented as part of year ending March 2022 evaluation of Violence Reduction Units (VRUs). The annex also provides additional outputs from the synthetic control group analysis.

Data sources

This section details the outcomes data accessed for the evaluation. This includes the geographical coverage and level of data, and key decisions taken when preparing data for analysis.

Hospital admissions data

Hospital admissions represent the most serious non-fatal violence as it is a violent injury that could not be treated in an emergency department. Approximately 30% of attendances for violent injury are admitted (Quigg et al., 2012). When a patient is admitted, the reason for admittance is recorded using the ICD-10 set of indicators. These appear in the NHS Digital records in several forms – ‘finished admissions episode’ is the most suitable way to identify these admissions. The admission is recorded against the month in which it concluded (as opposed to the date of admission or the incident that led to the admission). Therefore, it may not capture the small number of prolonged admissions that may have begun but not concluded during the data collection period and some admissions may have begun before the introduction of funding but concluded afterwards.

In the NHS Digital data set, admissions are attributed to hospitals located within a police force area (PFA) and local authority (LA) and the month in which the admission concluded is recorded. Accordingly, admissions for violence in those hospitals are attributed to that PFA/LA, which allows the data to be used to examine differences in admission for violence in VRU and non-VRU areas and the timing of the admission (before or after SV funding) allows pre-post trends and a synthetic control group approach to be employed.

The analysis incorporated all hospital admissions for violent injury (ICD-10: X92-Y09) and a sub-group, hospital admission for violent injury with a sharp object (ICD-10: X99).

Hospital admissions for any injury incurred in violence were not a primary outcome of the VRU programme, but given the range of VRU activities, it is theoretically plausible and very likely that the VRU activity could have impacted all violence. There are 16 categories of violent admissions in the ICD-10 catalogue. This includes varied mechanisms such as assault by bodily force and assault by different types of firearm, as well as assault by drowning, assault by smoke, fire and flames, and 2 groups of ‘Other’ (specified and unspecified) assault mechanisms. The team considered limiting the categories to assault mechanisms that could reasonably fit within a theoretical framework of VRU activity. However, because the range of activities by VRUs tend to aim to affect behaviour rather than a mechanism of violence, the team could not confidently limit the data via theory. The team also considered using objective statistical techniques, identifying the point of maximum curvature using the data set of admissions before SV funding (2012 to 2018), to limit the data set to the most common types of violence. Using this analysis, the team identified a set of 6 admission injury mechanisms, which included ‘Other unspecified’ but excluded theoretically relevant mechanisms such as handguns and rifles. This was unsatisfying as it was not theoretically sensible. In addition, such a statistical approach would create limits for the generalisability of findings beyond England and Wales. Accordingly, the team opted to include all admissions for violent injury within the ICD-10 catalogue.

Homicide data

The Home Office provided homicides data at the PFA level. The data sources were:

• quarterly Homicide Index (HI) data covering April to June 2016 through to July to September 2022
• monthly police recorded crime (PRC) homicides covering August 2013 to November 2021

The HI is continually updated (as police investigations and court cases progress) and is the primary data source for national publications on homicides. These data were used for the descriptive statistics and the programme-level synthetic control group analysis (using quarterly data) presented.

Recognising the low number of quarterly homicides, the team also conducted synthetic control group analysis on annual homicides. To meet the minimum number of pre-intervention periods required for the analysis, PRC monthly homicides data was aggregated to annual homicides.

To avoid biasing the impact estimates, homicides related to the following events were removed from both data sources:

• in year ending March 2017, 96 victims of the Hillsborough disaster and 4 victims of the Westminster Bridge attack
• in year ending March 2018, 31 victims of the terrorist attacks that involved multiple victims, including the Manchester Arena bombing and the London Bridge attack, and 11 victims of the Shoreham air crash
• in year ending March 2020, 39 victims whose bodies were found in a lorry in Essex

Police recorded crime data

Police recorded violence monthly data were provided by the Home Office at the Community Safety Partnership (CSP) level. Each PFA comprises multiple CSPs, which are broadly aligned to LAs. At the time of data collection, all police forces (excluding Greater Manchester) submit these data to the Home Office. Monthly counts for the following offence types were provided:

• violence with injury
• violence without injury
• possession of weapons offences

Data were also provided at the Lower Super Output Area (LSOA) level to facilitate exploratory analysis to understand the impact of SV funding in sub-areas with the highest (historical) levels of violence.

Constructing the counterfactual

To estimate the impact of SV funding, a counterfactual (what would have likely happened in the absence of funding) was constructed through:

• where available data allowed, conducting analysis at the LA or CSP-level, where treated (VRU) and untreated (non-VRU) areas are more comparable in population size and outcomes of interest
• population adjusting outcomes data to make treated (VRU) and untreated (non-VRU) areas to improve comparability
• forming synthetic control groups (a weighted average of CSPs in non-VRU areas) that followed a similar pre-SV funding trend in (population adjusted) PRC data to CSPs in VRU areas

The first 2 points are discussed and illustrated below using PRC data. Focusing on LA-level data and population adjustment improved the comparability of hospital admissions data. Homicides data were only available at PFA-level but population adjustment could still improve comparability.

Analysis at CSP-level

The team selected PFAs for SV funding based on levels of SV. The 18 PFAs experiencing the highest levels of SV received SV funding and established VRUs. However, these higher levels of violence are driven by the larger populations in VRU areas.

Funded PFAs typically cover large geographical areas and/or have more densely populated urban centres (relative to non-funded PFAs). The ‘box and whisker’ diagram below (Figure B.1) shows that the population distributions in funded (VRU) and non-funded (non-VRU) PFAs are not very comparable (there is a limited overlap between the distributions).

Figure B.1: Distribution of police force area populations (excludes outlier Metropolitan Police Service (population 8.9 million))

However, using smaller and more similar sized funded and non-funded geographical units (in this case, CSPs) improves comparability. This is evident in Figure B.2 where there is greater overlap between VRU and non-VRU CSP population distributions than observed in Figure B.1. It is still worth noting that CSP populations within funded areas were, on average, larger (reflecting a concentration of cities/large towns) than those in non-funded areas.

Figure B.2: Distribution of Community Safety Partnership populations

Population adjustment

Figure B.3 shows how population adjusting the outcomes can further support the construction of an appropriate counterfactual. The left-hand panel shows the average monthly count of police recorded violence with injury offences in funded and non-funded CSPs. Note that the left-hand panel very closely mirrors Figure B.2, highlighting the association between population size and outcomes of interest. The right-hand panel shows the same outcome but after population adjustment (average monthly rate per 100,000 persons), where comparability between funded and non-funded CSPs is substantially improved.

Figure B.3: Comparison of average violence and violence rates in VRU and non-VRU CSPs

Synthetic control method

Synthetic control methods (SCMs) seek to construct an appropriate counterfactual by creating a synthetic control group, which is a weighted average of potential comparator areas. Comparator (non-VRU) areas (PFAs/LAs/CSPs) that are more similar to the treated (VRU) areas on pre-intervention (population adjusted) outcome trends receive a heavier weighting than those that are less similar.

As well as providing reliable impact estimates (see O’Neill et al., 2020), SCMs can provide overall/average impact estimates for the entire treatment period, and average/cumulative impact estimates for each individual treatment period (for example, months, quarters). The latter is particularly useful for VRUs where it is anticipated impacts will be gradual and accumulate over time.

The precise approach was generalised synthetic control groups (see Xu, 2017), which allows for the inclusion of multiple treated areas (and constructing a synthetic control group for each). Diagnostic plots were assessed to ensure there was sufficient overlap between treated and comparator areas for impact estimates to be considered reliable, this is referred to as the ‘common support’ assumption.

Estimating VRUs’ impact (independent of Grip activity)

Additional analysis to, as far as possible, estimate the impact of VRUs independent of Grip (formerly Surge) activity was conducted on PRC. SV funding covers (more preventative) VRU activity and Grip activity, which is police enforcement focused. Specifically, Grip/Surge activity has a large focus on ‘hotspot’ policing.

To understand the degree to which hotspot policing contributes to the impacts observed across the wider VRU areas, the team undertook analysis on neighbourhoods (LSOAs) with the highest (pre-SV funding) levels of violence (‘hotspots’). Whilst some VRU activity is targeted (geographically and/or at high-risk individuals/groups), it is less targeted than Grip and impacts are anticipated to be observable across the wider VRU area.

This approach is presented as exploratory. Whilst it can be considered a reasonable approximation of Grip/Surge activity, it does not capture all activity and the effect this may have had on police recorded violence. Furthermore, reliable LSOA-level data was not available for all PFAs.

Assumptions and limitations

A key consideration for the synthetic control group analysis was the impact of COVID-19 on violence. Specifically, the impact of lockdowns/restrictions on violence outcomes. The team theorised that restrictions on people’s movement reduced the opportunity for violence. To test this theory, the team examined Google mobility data.

Figure B.4 shows that, on average, LAs in VRU and non-VRU areas followed a very similar trend in residential movement (time spent at home) between February 2020 and December 2021. Regression analysis, which tested the interaction between VRU status and time on movement, confirmed there was no statistically significant difference between LAs in VRU and non-VRU areas (estimate = 0.00009, p-value = 0.89).

Figure B.4: Percentage change in residential movement in VRU and non-VRU areas

Recognising the above, the impact estimates (which are the differences between VRU areas and the synthetic control group constructed from non-VRU areas) can be considered net of any effects on violence resulting from changes in movement related to COVID-19. In other words, it was not necessary to adjust the synthetic control group analysis for changes in movement because VRU and non-VRU areas experienced very similar (and not statistically significantly different) trends in movement. However, it will be important for police recorded violence trends (and impact estimates) to be continually monitored as more data not affected by COVID-19 restrictions becomes available.

Whilst the above showed the differences in movement between LAs in VRU and non-VRU areas were, on average, similarly affected by COVID-19, Google mobility data was not available at more granular levels to test this at the LSOA level. As such, some caution is advised with the analysis of ‘hotspots’, which used LSOA-level PRC data. It is possible that COVID-19 affected movement in LSOAs differently.

Another consideration was the level of analysis conducted to understand changes in violence trends. Whilst monthly data were provided, the main analysis presented included aggregating the data to fiscal quarters. The team decided to better capture underlying trends in violence, which were less clear (owing to volatility) in the monthly data.

It is important to note the 2 main potential limitations of the police recorded violence data:

1. Not all violence is reported to the police by the public and this is not captured in the data. Similarly, the police do not detect all violence. Increased levels of reporting and/or detection do not necessarily reflect a real increase in the levels of violence an area is experiencing. It is possible that VRU activity (for example, building relationships with communities affected by violence, better use of data) and Grip activity (additional enforcement) could affect the reporting and/or detection of violence.
2. Changes in the way the police record violence. This can include changes to and/or improvements over time in data capture systems.

Whilst results from the police recorded violence data should be interpreted with the above points in mind, the team conducted analysis at the whole programme level, which can be considered less affected by potential changes in the reporting, detecting or recording of violence in specific PFAs (they are averaged out against other PFAs).

There are some limitations to using hospital admissions data as an indicator of serious violence (SV) in a PFA. Firstly, the catchment areas of emergency departments are not necessarily co-terminus with PFAs. As patients typically travel to their closest emergency department for treatment (Haas et al., 2015), this may not be in the PFA where the violent incident occurred. Similarly, patients injured in more rural areas – which are also likely to have lower levels of violence and therefore less likely to receive SV funding – have to travel to population centres such as towns and cities for treatment meaning that violence captured by the hospital(s) in a PFA may not have occurred in that area. Hospital admissions are also susceptible to resourcing issues that may affect their validity. For example, hospitals that have fewer resources and staff may be less likely to admit a patient for a violent injury than a hospital with more resources. In addition, the availability of a minor injury treatment centre within the hospital’s catchment area may reduce the number of patients attending the emergency department (Rudge et al., 2013).

Programme-level synthetic control group analysis

Hospital admissions- sharp object violence

The team conducted synthetic control group analysis to estimate the impact of SV funding on sharp object violence admissions rates across all VRU areas.

Figure B.5 shows the average quarterly admissions rate (per 1 million persons) for LAs within VRU areas and the synthetic control group constructed. As described in the methodology section, the synthetic control group is a weighted average of non-VRU LAs. Relative to the (unweighted) average admissions rate in non-VRU areas, the synthetic control group provides a better counterfactual. Drawing on the results from the synthetic control group analysis, Figure B.6 shows the cumulative impact (following SV funding).

There was no statistically significant (at the 95% level) impact on quarterly admissions for sharp object violence at the LA level. The estimated average quarterly impact of SV funding on sharp object admissions was -0.02 (95% CI [-0.10 , 0.07]).

The estimated cumulative impact after 11 quarters of SV funding was -0.12 (95% CI [-1.20, 0.81]).

Figure B.5: Quarterly admissions for sharp object violence – synthetic control group analysis at local authority level

Figure B.6: Quarterly cumulative effect of VRU funding on admissions for sharp object violence at local authority level

Hospital admissions- any violence

Figure B.7 shows the average quarterly admissions rate (per 1 million persons) for LAs within VRU areas and the synthetic control group constructed. Figure B.8 shows the cumulative impact (following SV funding).

There was no statistically significant (at the 95% level) impact on quarterly admissions for violence at the LA level. The estimated average quarterly impact of SV funding on admissions was -0.36 (95% CI [-0.79 , 0.08]).

The estimated cumulative impact after 11 quarters of SV funding was -3.74 (95% CI [-8.90 , 0.81]).

Figure B.7: Quarterly admissions for any violence – synthetic control group analysis at local authority level

Figure B.8: Quarterly cumulative effect of VRU funding on admissions for any violent injury at local authority level

Homicides

Figure B.9 shows the average quarterly homicide rate (per 1 million persons) for VRU areas and the synthetic control group constructed. Relative to the (unweighted) average homicide rate in non-VRU areas, it can be seen that the synthetic control group provides a better counterfactual. However, there was some divergence in 2018. Figure B.10 shows the cumulative impact on homicide rates.

There was no statistically significant (at the 95% level) impact of SV funding on population adjusted homicide rates. The estimated average (quarterly) impact of SV funding on homicides was -0.27 (95% CI [-0.65 , 0.11]) per 1 million persons.

Whilst not statistically significant, by quarter 2 (July to September), 2022 (the most recent data available) there was an estimated reduction of -2.84 (95% CI [-6.67, 1.05]) in homicides per 1 million persons.

Figure B.9: Quarterly homicides – synthetic control group analysis

Figure B.10: Cumulative impact on homicides

Police recorded violence

Figure B.11 shows the average quarterly police recorded violence with injury rate (per 1 million persons) for VRU areas and the synthetic control group constructed. Relative to the (unweighted) average violence with injury rate in non-VRU areas, the synthetic control group provides a better counterfactual. Following SV funding, differences in violence with injury rates between VRU areas and the synthetic control group appear limited.

Figure B.11: Violence with injury (fiscal quarters) – synthetic control group analysis

Figure B.12 shows the average quarterly violence without injury (per 100,000 persons) for VRU areas and the synthetic control group constructed. Relative to the (unweighted) average violence without injury rate in non-VRU areas, the synthetic control group provides a better counterfactual. Following SV funding, there is a clear divergence between the average violence without injury rate in VRU areas and the synthetic control group, which indicates a potential impact.

Figure B.12: Violence without injury (fiscal quarters) – synthetic control group analysis

Figure B.13 shows the average quarterly possession of weapons offences (per 100,000 persons) for VRU areas and the synthetic control group constructed. Relative to the (unweighted) average possession of weapons offences in non-VRU areas, the synthetic control group provides a better counterfactual here too. Similar to violence without injury offences, a divergence between the rates in VRU areas and the synthetic control group are apparent following SV funding.

Figure B.13: Possession of weapons offences (fiscal quarters) – synthetic control group analysis

Figure B.14 summarises the results from the synthetic control group analysis with cumulative impacts. By fiscal Q3 2022 (October to December 2021), the estimated cumulative impacts of SV funding on police recorded violence rates in VRU areas were:

• for violence with injury offences, a small (not statistically significant) estimated reduction of -2.05 (95% CI [-125.16, 76.11]) per 100,000 persons
• for violence without injury offences, an estimated (statistically significant) reduction of -243.22 (95% CI [-541.82, -51.58]) per 100,000 persons
• for possession of weapons offences, an estimated (statistically significant) reduction of -15.63 (95% CI [-28.1, -0.59]) per 100,000 persons

Figure B.14: Cumulative impacts on police recorded violence (fiscal quarters)

Hotspot analysis

Data coverage

The Home Office provided monthly police recorded offending data at the LSOA level for 33 PFAs. After data cleaning and sense checks, the team retained data for 18 PFAs for analysis. Sense checks included comparing PFA violence count totals from the LSOA-level data, where recording is more variable, to CSP-level data to understand data coverage and consistency over time. Table B.1 shows the 18 PFAs and their SV funding (or ‘treatment’) status.

Table B.1: SV funding status of police forces included in analysis

SV funded	Not funded
Avon & Somerset	Cambridgeshire
Bedfordshire	Devon & Cornwall
Hampshire	Durham
Merseyside	Gloucestershire
Metropolitan Police	Lincolnshire
Northumbria	Norfolk
Nottinghamshire	Northamptonshire
South Yorkshire	Surrey
Sussex	West Mercia

Data was aggregated to quarterly (from monthly) to better capture underlying trends and facilitate the synthetic control group approach (see next section ‘Defining hotspots’).

Defining hotspots

LSOAs in SV-funded PFAs were defined as hotpots based on the total violence (with and without injury) in the 3 years prior to SV funding. LSOAs were arranged (high to low) by their total violence and assigned to quantiles (in equally sized groups). Two definitions were operationalised and tested:

• LSOAs in the top decile (10%) of total violence, within each PFA
• LSOAs in the top centile (5%) of total violence, within each PFA

Recognising that PFAs cover large geographical areas and the focusing of Grip activity, attribution is likely stronger for analysis focused on the top 5% of LSOAs. Table B.2 shows how these 5% of LSOAs account for approximately a quarter of all violence with injury offences in their wider PFA (based on year ending March 2019 data).

Table B.2: Violence with injury offences in hotspots (top 5% of LSOAs)

Police force	Total offences	Offences in hotspots	Offences in hotspots (%)
Avon & Somerset	12,251	3,396	27.7
Bedfordshire	4,704	1,476	31.4
Hampshire	20,353	4,901	24.1
Merseyside	14,600	3,838	26.3
Metropolitan Police	76,941	18,371	23.9
Northumbria	13,237	3,634	27.5
Nottinghamshire	13,643	3,500	25.7
South Yorkshire	14,344	3,726	26.0
Sussex	13,869	3,895	28.1
Total	183,942	46,737	25.4

Results

There were statistically significant reductions in violence with injury and violence without injury offences after introducing SV funding in violence hotspots. Table B.3 shows that in the top 5% of LSOAs, there was an average (per LSOA) reduction of 33 violence with injury and 20 violence without injury offences over the SV funding period (April 2019 to December 2021), relative to the synthetic control group. The total reduction (across the 583 LSOAs defined as hotspots) was 19,420 violence with injury and 11,940 violence without injury offences.

Table B.3: SV-funded hotspot results (per LSOA)

Hotspot definition	Offence	Cumulative impact	Lower CI	Upper CI
Top 10%	Violence with injury	-20.36	-23.81	-17.35
	Violence without injury	-10.63	-16.71	-5.06
Top 5%	Violence with injury	-33.31	-38.75	-27.82
	Violence without injury	-20.48	-31.53	-8.87

Figure B.15 shows the trends in violence with injury and violence without injury offences for the SV-funded hotspots (both definitions) against the synthetic control groups. In all cases, trends in SV-funded hotspots started to diverge from the synthetic control groups (indicating a reduction) shortly after introducing funding and were (generally) maintained over the whole funding period.

There were substantial drops in violence during quarter 1, 2020 (April to June) and quarter 3 to quarter 4, 2020 (September to March) in SV-funded hotspots and the synthetic control groups, which align with COVID-19 national lockdowns. For violence with injury offences, SV-funded hotspots appeared to have greater reductions during the national lockdowns, relative to the synthetic control groups. Whilst this could be the result of Grip/Surge activity, it is also possible that SV-funded hotspots were more affected by the national lockdowns.

During/following lockdowns easing, the trends in SV-funded hotspots remained below the synthetic control group (and their pre-SV funding / COVID-19) levels. It is worth noting that Surge/Grip activity had a specific focus on managing/preventing anticipated spikes in violence following lockdowns.

Figure B.15: Violence trends in SV-funded hotspots and synthetic control groups

References

[Anderson, R. (2007) ‘Thematic content analysis (TCA)’. Descriptive presentation of qualitative data, pp.1-4.]

[Braga, A.A., Turchan, B., Papachristos, A.V. & Hureau, D.M. (2019) ‘Hot spots policing of small geographic areas effects on crime’. Campbell Systematic Reviews, vol. 15(3), p. e1046.]

Brophy, S., Bandyopadhyay, A., Kennedy, N., Jones, H., Harrington, K., Evans, J., Bellis, M., Rowe, B., Spasic, I., McNerney, C. & Moore, S.C. (2022) ‘Unlocking data to inform public health policy and practice’. Swansea University. [accessed 19/12/2022].

[Bullock, K., Agar, I., Ashby, M., Brennan, I., Hales, G., Sidebottom, A. & Tilley, N. (2022) ‘Police practitioner views on the challenges of analysing and responding to knife crime’. Preprinted ahead of publication.](https://osf.io/preprints/socarxiv/k2npr/0 [accessed 19/12/2022].

[Florence, C., Shepherd, J., Brennan, I. and Simon, T. (2011) ‘Effectiveness of anonymised information sharing and use in health service, police, and local government partnership for preventing violence related injury: experimental study and time series analysis’, British Medical Journal, vol. 342.]

Gaffney, H., Farrington, D., & White, H. (2021a) ‘Cognitive Behavioural Therapy’. [accessed 19/12/2022].

Gaffney, H., Jolliffe, D., & White, H. (2021b) ‘Emergency department violence interventions’. [accessed 19/12/2022].

Goldacre, B., & Morley, J. (2022) ‘Better, Broader, Safer: Using health data for research and analysis’. A review commissioned by the Secretary of State for Health and Social Care. Department of Health and Social Care. [accessed 19/12/2022].

[Haas, B., Doumouras, A.G., Gomez, D., de Mestral, C., Boyes, D.M., Morrison, L. et al. (2015)’Close to home: an analysis of the relationship between location of residence and location of injury’. Journal of Trauma and Acute Care Surgery, vol. 78(4), pp.860-865.]

Heeks, M., Reed, S., Tafsiri, M. & Prince, S. (2018) The economic and social costs of crime: Second edition, Home Office Research Report 99.Home Office (2020) ‘Violence Reduction Unit Interim Guidance’.

[O’Neill, S., Kreif, N., Sutton, M. and Grieve, R. (2020) ‘A comparison of methods for health policy evaluation with controlled pre‐post designs’. Health services research, vol. 55(2), pp.328-338.]

ONS (2022) ‘The nature of violent crime in England and Wales: Year ending March 2022’, Office for National Statistics. [accessed 19/12/2022]

[Quigg, Z., Hughes, K., and Bellis, M.A., 2012. (2012)’Data sharing for prevention: a case study in the development of comprehensive emergency department injury surveillance system ad its use in preventing violence and alcohol-related harms’. Injury prevention, vol. 18(5), pp.315-320.]

[Rudge, G. M., Mohammed, M. A., Fillingham, S. C., Girling, A., Sidhu, K., & Stevens, A. J. (2013). The combined influence of distance and neighbourhood deprivation on emergency department attendance in a large English population: a retrospective database study. PloS one, 8(7), e67943.]

[Walker, D., and Myrick, F. (2006) ‘Grounded theory: An exploration of process and procedure’. Qualitative health research, vol.16(4), pp.547-559.]

White, H., Saran, A., Verma, A., Oprea, E. & Babudu, P. (2021) ‘Evidence and Gap Map of Interventions to Prevent Children Getting Involved in Violence: Technical Report on the First Edition’, Youth Endowment Fund and Campbell Collaboration. [accessed 19/12/2022].

[Xu, Y. (2017) ‘Generalized synthetic control method: Causal inference with interactive fixed effects models’. Political Analysis, vol. 25(1), pp.57-76.]

YEF (2022) ‘What the toolkit tells you’, Youth Endowment Fund. [accessed 19/12/2022].

Footnotes

1. Total investment includes additional funding made available for Serious Youth Violence Interventions in year ending March 2022. ↩

2. These figures are based on the central estimate of 243 (per 100,000 persons) violence without injury offences prevented. The upper and lower estimates were 52 (£0.87 return on investment) and 542 (£9.14 return on investment) offences prevented (per 100,000 persons), respectively. ↩

3. Elements adapted from Home Office (2020). ↩

4. Hospital admissions was selected as a primary outcome measure by the Home Office in consultation with NHS Digital. Emergency Department data were considered but, owing to more limited coverage and quality, were not selected as a primary outcome measure or accessed for the evaluation. ↩

5. LSOAs are small geographical areas with populations ranging from around 1,000 to 3,000. ↩

6. SV-funded areas where reliable LSOA level data was available included: Avon & Somerset, Bedfordshire, Hampshire, Merseyside, Metropolitan Police, Northumbria, Nottinghamshire, South Yorkshire and Sussex. ↩

7. It should also be noted that over the same time period, there was a decline in violence in other parts of the UK, without the presence of a VRU. ↩

8. Trauma-informed training focuses on professionals recognising trauma and avoiding re-traumatisation. ↩

9. See Youth Endowment Fund (YEF) Toolkit. ↩

10. Two interventions (Thames Valley RAW and SOFEA) are based across a mixture of A&E departments, the community and custody. ↩

11. VRU Year 3 SoR Annex F – Violence Reduction Unit Funding Application Guidance 2021 to 2022. ↩

12. VRU Year 3 SoR Annex A – Additional Investment Pot Bid Criteria. ↩

13. Note that there was limited evidence of VRUs undertaking centralised co-design of interventions with young people (including Serious Violence Youth Programme interventions). This was more likely to be undertaken through delivery partner mechanisms for services that were pre-existing. See Chapter 5 on Community Engagement for more detail. ↩

14. Interventions might cover more than one of these priorities or themes. ↩

15. Early intervention delivery models accounted for 40% (79) of interventions in year ending March 2022 and 24% (51) of interventions in year ending March 2021. ↩

16. Early intervention delivery models targeted at level 2 - high risk groups accounted for 16% (32) of interventions in year ending March 2022 and 5% (11) of interventions in year ending March 2021. ↩

17. Prevention delivery models accounted for 19% (38) of interventions in year ending March 2022 and 32% (69) of interventions in year ending March 2021. ↩

18. Note that as the monitoring data from all VRUs does not indicate how many initiatives any individual young person took part in, there is the possibility of double counting. ↩

19. See Youth Endowment Fund (YEF) Toolkit. ↩

20. Note that number of people reached by target group figures are broken down by the predominant target group reported by VRUs. ↩

21. Most interventions had not yet completed their delivery cycle and limited and/or they had made inconsistent progress in relation to the monitoring and evaluation of these. The evaluation also could not establish whether interventions had met their objectives, as intended targets were not set out at the outset (or they were not provided to the national evaluation team). ↩

22. Outcome data for Serious Violence Youth Intervention Programmes was out of scope for the national evaluation. ↩

23. Due to limitations experienced as a result of the COVID-19 restrictions, the young people element of the evaluation was adapted. ↩

24. VRU Year 3 SoR Annex A – Additional Investment Pot Bid Criteria. ↩

25. Data based on information drawn from quarter 4 (January to April 2022) data from 17 out of 18 VRUs. ↩

26. Data from year ending March 2021 and year ending March 2022 based on information drawn from January to April 2022 data from 17 out of 18 VRUs; however, the omitted VRU differed across the 2 years. ↩

27. Data from year ending March 2021 and year ending March 2022 based on information drawn from January to April 2022 data from 17 out of 18 VRUs; however, the omitted VRU differed across the 2 years. ↩