Research and analysis

Violence reduction unit year ending March 2021 evaluation report

Published 1 April 2022

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. VRUs take a preventative, whole-system approach to violence reduction, which comprises:

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

The Serious Violence Fund also covers Surge activity, which is enforcement focused. Combined, VRU and Surge funding represents a total investment of £174m across the financial years 2019 to 2020 and 2020 to 2021 to tackle serious violence (SV). The 18 areas were selected, and funding amounts allocated, based on levels of SV between the financial year ending 2016 to that ending 2018.

The Home Office commissioned Ecorys, Ipsos MORI, the University of Hull and the University of Exeter to conduct a process and impact evaluation of VRUs in their second year of operation (April 2020 to March 2021). The broad aims of the evaluation were to:

• undertake a process evaluation, and linked theory-based impact evaluation, to understand how VRUs are implementing a whole-system approach to violence reduction in their second year of (Home Office) funding and how this is contributing to VRUs meeting their aims
• estimate the impact of VRUs against outcomes specified by the Home Office, and others also related to violence reduction.

The process evaluation (and linked theory-based impact evaluation) included a comprehensive document review to develop a theory of change (ToC) for each VRU and, based on these, the programme-level analytical framework and research tools. In-depth qualitative research was subsequently undertaken, with all 18 VRUs capturing the views and experiences of 290 stakeholders/beneficiaries in total.

The impact evaluation applied multiple quasi-experimental designs, which estimate what would have likely happened in the absence of funding, on a range of (serious) violence outcome measures to estimate as robustly as possible the initial impacts of the Serious Violence Fund.

Key findings

There are early indications of VRUs impacting on violence reduction. While there was no (statistically significant) impact detected on the most serious SV outcomes (hospital admissions and homicides), there were impacts on police recorded violence. Between April 2019 (when SV funding was deployed) and September 2020, it is estimated that 41,377 violence without injury offences had been prevented in funded areas, relative to non-funded areas. Alongside a reduction of 7,636 violence with injury offences, this represents potential costs avoided of £385m. Relative to the amount of SV funding over the same period, this represents a return of £3.16 for every £1 invested ^{[footnote 1]}. Recognising VRUs are a longer-term preventative approach, these results are encouraging. Past research into public health approaches to preventing violence (e.g. the Scottish VRU and the Cardiff model) indicates a greater impact will accumulate over the medium to longer term for VRUs. Furthermore, VRU and Surge activity appear to be complementing each other with a greater combined impact (and positive working relationships reported by stakeholders).

In terms of different elements of the whole-system approach, VRUs have continued to make progress since year 1. This was particularly evident for multi-agency working and data sharing, where previous work provided a strong foundation to build on. While some progress was evident for engaging young people and communities, and commissioning/delivering interventions, these were generally more susceptible to challenges resulting from the COVID-19 pandemic. The short-term nature of funding was also recognised as a limiting factor to embed a longer-term whole-system approach. Key findings against each element of the whole-system approach included:

• multi-agency working:
  • VRUs have made good progress in developing partners’ understanding, skills and practices of preventative approaches to SV
  • at the strategic level, VRUs continued to strengthen partnerships, streamline approaches, and identify and facilitate areas for collaboration; owing to the large geographical coverage of the VRUs, some reported slower progress engaging (smaller) community and voluntary sector organisations
  • multi-agency working was more developed at the strategic level than at the operational level; the latter was recognised as something that takes longer to achieve; VRUs, where this was more developed at the operational level (e.g. those building on VRUs that were in place prior to SV funding), reported that ensuring strategic alignment was essential for the vision of the VRU to take effect at the operational level
• data sharing and analysis:
  • the Strategic Needs Assessments (SNAs) – which identify local trends in violence, crime and their underlying causes – developed in year 1 provided an opportunity for VRUs to reflect on and refine their data focus; this has led to a greater appreciation for data as a tool to prevent violence
  • police data continued to be the most accessible and used source by VRUs; data are most commonly accessed and analysed at an aggregate level (i.e. overall numbers in sub-areas and/or for specific cohorts); some VRUs are accessing individual-level data; progress has been made in accessing health, education and other sources but some challenges, particularly the requirement of multiple/complex data sharing agreements, persist
  • having the expertise to analyse data from multiple sources in a meaningful and insightful way was identified as a key enabler; some VRUs were working with external experts, with experience of the data and the required data sharing agreements, to provide analysis in an accessible format (e.g. data dashboards)
  • VRUs had/were developing overarching outcomes frameworks, which they intended to track progress against; however, recognising the scale and range of interventions commisioned, there was some challenges in collecting consistent and accurate data at this level
• young people and community engagement:
  • engagement activities were designed and delivered to serve a variety of purposes; some VRUs focused on understanding experiences of crime and SV, some aimed to understand communities’ needs and concerns, while others planned to embed young people and communities in the design of the interventions they commissioned
  • despite most VRUs having engagement plans in place in year 2, only limited progress had been made to engage communities and young people in activities that would inform their strategy and objectives; a significant challenge in year 2 was the impact of COVID-19 restrictions; moving into year 3, young person and community engagement is a priority for VRUs
  • VRUs can learn from and/or link in to VRU partners’ existing engagement strategies
• commissioning and delivering evidence-based interventions:
  • there were signs of VRUs making progress in terms of strategically commissioning interventions; this was most evident in the commissioning of interventions which aligned to the SNAs developed by VRUs; there was some evidence of the comissioned interventions being evidenced-based, but it was too early to provide a comprehensive assessment owing to missing data, funding cycles ongoing and the impact of COVID-19
  • with an estimated 262,000 individuals supported in year 2, there is early indicative evidence to suggest that VRUs had effectively reached children, young people and adults through the delivery of interventions; recognising interventions were ongoing, limited progress in monitoring/evaluation, and delivery being affected by COVID-19 restrictions (as was the research with young people), the effectiveness of the individual interventions could not be assessed at this stage; however, young people who could take part in the research reported positive perceptions and benefits of their interventions

Recommendations

In terms of the future development of VRUs, the evaluation findings support the following recommendations:

For the Home Office to:

• support VRUs to improve the quality of intervention monitoring data which informs, and underpins, evidence-based commissioning
• urge other central government departments, including Department of Health and Social Care (DHSC) and Department of Education (DfE), to improve data access at VRU level and ensure partners take responsibility for their role and involvement in the VRU

And for VRUs to:

• ensure clear, evidence-based rationale for commissioned interventions, supported by the Youth Endowment Fund (YEF) toolkit
• increase focus on effective engagement of voluntary and community sector at a strategic level
• consider the potential role of external experts to navigate common data-sharing challenges and increase data analysis and insight capacity
• share any good practice toolkits or resources on implementing whole-system approach with all VRUs through learning networks

Acknowledgements

The report authors wish to thank all individuals who contributed to this evaluation. In particular, we acknowledge the support and participation of all 18 VRU teams and their local stakeholders/beneficiaries. We also thank Home Office analysts and policy colleagues, for their support throughout the evaluation, and the wider evaluation team at Ecorys and Ipsos MORI for all their contributions to the fieldwork and analysis.

Special thanks also to the University of Exeter for providing additional funding through the Strategic Priorities Fund to support the evaluation.

1. Introduction

This report presents findings from the process and impact evaluation of Violence Reduction Units (VRUs) in the financial year 2020 to 2021, the second year of their operation. VRUs have been funded 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 SV and its root causes. This VRU evaluation was commissioned by the Home Office in July 2020 and conducted by Ecorys in partnership with Ipsos MORI, Prof. Iain Brennan (of the University of Hull) and Prof. Mark Kelson (of the University of Exeter).

This chapter begins with an overview of the policy background and context for the establishment of VRUs, followed by a description of the evaluation aims, objectives and methodology. The chapter then provides the overall theory of change (ToC) for VRUs that underpinned/informed the evaluation design and implementation. The chapter concludes by setting out the scope and structure of the remainder of the report.

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 areas were selected by their levels of SV experienced between the financial year ending 2016 and that ending 2018. In early 2020, funding was confirmed for a second year (April 2020 to March 2021).

The core aim of a VRU is to provide leadership and strategic co-ordination of all relevant agencies, to support a public health/whole-system approach to tackle SV and its root causes. Alongside the VRU core function, each PFA is required to fund specific interventions working with young people (aged under 25).

Key aspects of the whole-system approach, as summarised in Home Office (2020), include:

• working with and for communities (and young people)
• unconstrained by organisation or professional boundaries (i.e. multi-agency)
• focusing on a defined population
• a response comprising short- and long-term solutions

Additional hallmarks of the whole-system approach are the need for data and intelligence to understand the pattern of SV experienced by the population and evidence of effective interventions to respond to the problem. These reflect both the internal monitoring and reporting required of VRUs and the evaluation work undertaken. The key outputs that inform each VRU’s response are the:

• Strategic Needs Assessments (SNA), which explore trends in SV and associated risk factors to understand the at-risk cohort (i.e. target population) and local issues
• Response Strategies, which set out how the VRUs are seeking to prevent/reduce violence, in the context of the SNA

The Home Office specified three outcomes against which the VRUs would be measured:

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 evaluation covers the above and additional outcomes (both quantitative and qualitative).

The Serious Violence Fund also covers Surge activity. The same 18 PFAs received Surge funding, which, like VRUs, aims to reduce violence (measured against the same outcomes) but focuses on enforcement (e.g. increased policing of hotspot areas, Stop and Search), rather than the more preventative focus of VRUs.

Table 1.1 details the funding allocations by PFA. Key points to note about funding are:

• both VRU and Surge funding were allocated based on the volume of hospital admissions for assault with a sharp object between the financial year ending 2016 and that ending 2018
• VRU funding (£70m across two years) followed a tiered system, whereas Surge funding (more) directly reflected each force’s percentage share of overall hospital admissions
• each VRU received the same amount of money in the financial year 2020 to 2021 as they did in 2019 to 2020
• Surge funding accounted for more than VRU funding in both 2019 to 2020 (£63m) and 2020 to 2021 (£42m)
• some VRUs were subject to local match funding (e.g. additional funding from the Office of the Police and Crime Commissioner and/or in-kind resource from partner agencies)

Table 1.1: Serious Violence (SV) funding allocations (overall total in the financial years 2019 to 2020 and 2020 to 2021)

Police Force area	Surge	VRU	Total SV funding
London	£34,487,455	£14,000,000	£48,487,455
West Midlands	£12,601,485	£6,740,000	£19,341,485
Greater Manchester	£7,943,375	£6,740,000	£14,683,375
Merseyside	£6,955,100	£6,740,000	£13,695,100
West Yorkshire	£6,655,315	£6,740,000	£13,395,315
South Yorkshire	£4,269,410	£3,200,000	£7,469,410
Northumbria	£3,844,185	£3,200,000	£7,044,185
Thames Valley	£3,203,960	£2,320,000	£5,523,960
Lancashire	£3,009,610	£2,320,000	£5,329,610
Essex	£2,912,435	£2,320,000	£5,232,435
Avon & Somerset	£2,843,520	£2,320,000	£5,163,520
Kent	£2,742,215	£2,320,000	£5,062,215
Nottinghamshire	£2,543,730	£1,760,000	£4,303,730
Leicestershire	£2,316,990	£1,760,000	£4,076,990
Bedfordshire	£2,288,730	£1,760,000	£4,048,730
Sussex	£2,211,555	£1,760,000	£3,971,555
Hampshire	£2,090,250	£1,760,000	£3,850,250
South Wales	£1,980,680	£1,760,000	£3,740,680
Total	£104,900,000	£69,520,000	£174,420,000

1.2 Evaluation aims and objectives

The Home Office commissioned Ecorys, Ipsos MORI, the University of Hull and the University of Exeter to conduct a process and impact evaluation of VRUs. The 2020 to 2021 evaluation (covering year 2 of VRU funding and activity) follows research undertaken in 2019 to 2020 (year 1), which explored the early implementation of VRUs (process evaluation) and assessed the feasibility of a future impact evaluation. The broad aims of the 2020 to 2021 evaluation are to:

• undertake a process evaluation to understand how VRUs are implementing a whole-system approach to violence reduction in their second year of (Home Office) funding
• using a combination of approaches, estimate the impact of VRUs against outcomes specified by the Home Office, and others related to violence reduction

Specifically, the process evaluation explores how the VRUs are working with partners to deliver, and the progress made towards, core elements of the whole-system approach:

• multi-agency working
• data sharing and analysis
• involving young people and communities
• commissioning and supporting interventions

Each of these elements (or activities), and how they are anticipated to contribute to meeting the VRU aims, are detailed in the VRU programme-level ToC (set out in Section 1.3 below).

The impact evaluation seeks to explore the overall impact of SV funding (which includes enforcement-focused Surge activity, as well as VRUs) on key outcomes and, as far as possible, isolate the impact of VRUs (independent from Surge). Linked to the exploration of core elements of the whole-system approach (detailed above), the impact evaluation also considers how, and under what circumstances, outcomes/impacts occur.

1.3 VRU Theory of Change

Figure 1.1 depicts a (whole) programme-level ToC for VRUs.

The ToC sets out the intended outcomes/impacts of VRUs and the presumed causal pathways (indicated by arrows) – through inputs, activities and outputs – to reach these.

The activities, which are the key elements of a whole-system approach, enable the intended outputs; these in turn are anticipated to result in outcomes. Outcomes at the strategic and operational, and young people and community levels, facilitate the ultimate impacts of a sustainable whole-system approach, reduced violence, improved life outcomes for young people, and communities feeling safe.

Figure 1.1: Programme-level Theory of Change (ToC) for VRUs

1.4 Evaluation methodology

The evaluation employs a mixed-methods design to gather and analyse evidence through several approaches to meet the evaluation aims. The different elements/stages of the evaluation methodology are:

• document review: key documents for each VRU (2020 to 2021 funding applications, SNAs and Response Strategies) were reviewed to identify potential changes to VRUs since the 2019 to 2020 evaluation and to develop/revise VRU-level ToCs and outcome frameworks; VRUs’ annual and quarterly monitoring reports were also reviewed for inclusion in later stages of the evaluation and analysis
• development of programme-level analytical frameworks: based on the document review and initial consultations with VRU leads, an evaluation framework (see Annex 2) and VRU programme-level ToC were developed to guide subsequent evaluation activity including research instruments, information and engagement materials and analysis tools
• three phases of fieldwork to conduct qualitative research: a programme of in-depth consultations in all 18 VRU areas explored the implementation (process evaluation) and perceived effectiveness/impact (theory-based impact) of the core elements of the whole-system approach
  • the three phases of fieldwork involved 290 interviews and discussion groups across the 18 VRUs between August 2020 and March 2021; comprising:
  • 88 interviews with VRU leads, strategic VRU stakeholders and external partners
  • 69 interviews with VRU operational and analyst staff
  • 103 interviews with external operational and frontline staff (e.g. delivering VRU-funded interventions)
  • 30 interviews with young people supported by VRU-funded interventions
• Impact evaluation using quasi-experimental designs, alongside a theory-based impact evaluation

An overview of the methods employed and key limitations, which are intended to aid interpretation of the findings presented, are provided in the sections that follow. Additional detail for the impact evaluation is provided in Annex 3.

Process and theory-based impact evaluation methodology

For the process and theory-based impact evaluation, semi-structured topic guides aligned to the evaluation framework and programme-level ToC were developed to inform data collection via in-depth consultations. The theory-based impact evaluation harnessed all evidence generated through the evaluation (quantitative and qualitative) to assess the contribution of different elements of the whole-system approach to violence reduction and wider VRU aims.

Impact evaluation methodology – outcomes of interest

The impact evaluation focused on select outcomes where there is a strong theoretical link to VRU (and wider SV funding) activity and data requirements were met ^{[footnote 2]}:

• NHS held data on hospital admissions resulting from intentional injury caused by a sharp object for:
  • all ages
  • those under 25 years old
• Home Office held data on homicides
• Police recorded crime for the following offence groups:
  • Violence with injury
  • Violence without injury
  • Possession of weapons offences

As far as available and reliable data allowed, these outcomes are aligned to the Home Office key outcomes for VRUs (see section 1.2). While police recorded crime is not a key outcome for the Home Office, it was included as an outcome of interest as it captures all types of violence (not just knife enabled), which is consistent with the preventative nature of VRUs, and is routinely used by the VRUs to monitor their own progress.

Impact evaluation methodology – analytical approach

Quasi-experimental designs (QEDs) were employed to meet the aims of the impact evaluation (see section 1.2). SV funding was awarded to the 18 PFAs experiencing the highest levels of violence, specifically, the number of hospital admissions resulting from intentional injury caused by a sharp object between the financial year ending 2016 and that ending 2018. As such, simple comparisons to areas that did not receive funding (which were experiencing less violence) are not appropriate. When funding is not allocated at random, QEDs are required to construct a suitable counterfactual (which is an estimation of what would have likely happened in the absence of treatment). QED approaches included:

• Synthetic control methods (SCMs) to compare changes in outcomes following the introduction of SV funding between funded (‘treated’) and non-funded (‘comparator’) areas ^{[footnote 3]}; SCMs create a counterfactual by generating a weighted average of comparator areas based on their similarity on pre-VRU offending trends to treated areas; comparator areas, which depending on data availability, were PFAs and/or sub-areas within (e.g. community safety partnerships), that are more similar to VRU areas receive a heavier weighting than those that are less similar

• Interrupted time series analysis (ITSA) to understand how trends within treated areas are affected following the introduction of SV funding; ITSA creates a counterfactual by forecasting outcome trends based on pre-SV outcomes, which can then be compared to the observed (post-SV funding) trends

The primary approach was SCMs recognising these include comparisons between funded and non-funded areas ^{[footnote 4]}. ITSA was a secondary/supporting approach. Both approaches included consideration of the impact of the COVID-19 pandemic on the outcomes of interest ^{[footnote 5]}. Specifically, it was theorised that national lockdowns (and wider restrictions) affected people’s movement (e.g. they were spending more time at home) and, as such, this impacted on their exposure to / opportunity for violence. Exploratory analysis and models, using available data (e.g. data collected by Google on changes in movement), were conducted based on this theory.

To, as far as possible, isolate the impact of VRUs independent from Surge activity, a combined VRU and Surge ToC was developed (see Annex 4). This led to the inclusion of Stop and Search rates (a proxy for increased police activity) in exploratory SCMs, where resulting impact estimates were compared to those from SCMs without Stop and Search rates included.

The term ‘impact’ is used when reporting the results. This is in recognition that SV funding was a substantial investment and, as such, there is an argument for attribution. However, it should be noted that there are other (unobservable) external factors and potential activity linked to violence prevention not fully captured in the methods described above.

Limitations to the evaluation findings

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 the approach which are important to bear in mind when considering the evaluation findings in this report. Key limitations to the data collection and analysis were:

• difficulties accessing some operational and frontline stakeholders due to impact of COVID-19: the second phase of evaluation fieldwork focused on expanding the range of stakeholder perspectives to conduct consultations with core VRU staff, alongside operational and frontline stakeholders; however, a second national lockdown at the start of this phase of fieldwork reduced the availability and ability of staff in these groups to participate in the research, particularly as some key staff were again redeployed to deal with their organisation/service’s responses to the pandemic; as a result, the proportion of operational and frontline staff is lower than anticipated; this presented some challenges in triangulating/comparing their views with those of staff in more strategic and senior leadership roles
• identification, selection and engagement of young people in the research: it was necessary to negotiate access to young people through VRU operational leads, as well as staff in frontline providers and delivery organisations; while research teams took all practical steps to avoid potential biases, there were limitations in their ability to target and sample research participants (e.g. only young people engaged by projects could be interviewed)
• timeframe required for observing outcomes and impacts: the ‘whole-system approach’ adopted by VRUs is a longer-term investment in addressing the root causes of SV and, as reflected in the programme-level VRU ToC, some of the intended outcomes of VRUs are only likely to be observable in the longer term, beyond the timescales of this evaluation
• variation in quality and coverage of monitoring and evaluation data on VRU-funded interventions: the available data on VRU-funded interventions, including that collected as part of the quarterly reporting process, varied significantly in the mode of collection and the extent to which the outcomes presented could be verified; as a result, it was challenging to consistently estimate the scale and nature of VRU-funded interventions with young people, or to assess outcomes and ‘distance travelled’ by young people given the lack of consistent and reliable baseline and follow-up data

1.5 About this report

The structure of this final report is:

• Chapter 2 sets out the approach to, and examines results from, the QED exploring the impact of VRUs on reducing SV
• Chapter 3 outlines VRUs’ definitions of SV and the at-risk cohort(s), and the processes and evidence used to agree these definitions, alongside a review of any changes to VRU models of working since year 1
• Chapter 4 examines the multi-agency element of the whole-system approach to violence reduction; this includes the different approaches taken to multi-agency working, the extent to which VRUs’ multi-agency activity is being implemented effectively, and an assessment of the extent to which multi-agency working is contributing to VRU aims (i.e. theory-based impact evaluation)
• Chapter 5 focuses on the data and intelligence-sharing element of the whole-system approach; this includes progress since year 1 on approaches to data sharing and analysis, the extent to which VRUs are effectively implementing these, and an assessment of the extent to which data and intelligence sharing is contributing to VRU aims
• Chapter 6 examines the changes in approaches taken to young person and community engagement since year 1, the effectiveness of these approaches, and the extent to which young person and community engagement has contributed to VRU aims
• Chapter 7 provides an outline of approaches to commissioning VRU-supported interventions, the types/numbers of interventions, and the extent to which interventions are being effectively commissioned and delivered; this is followed by an assessment of the extent to which VRU interventions are contributing to VRU aims
• Chapter 8 presents conclusions on the extent to which VRUs have contributed to violence reduction and wider aims, recommendations arising, and a brief review of any areas for further investigation

2. Impact of VRUs on violence

This chapter presents results from the impact evaluation. Section 1.4 explains the methodological approach (with Annex 3 providing additional detail).

2.1 Impact on hospital admissions

The impact evaluation did not detect a statistically significant association (impact) between SV funding (the intervention) and hospital admissions resulting from intentional injury caused by a sharp object (the outcome). In any statistics model, there is some uncertainty around the (impact) estimates due to variability in the data. Confidence intervals (CI) report this uncertainty by providing a range (upper and lower estimates), where there is a high degree of confidence (95% probability) that the true value (impact) falls within. If the CI range includes zero (i.e. where the true value could be positive or negative), the impact estimate is not considered statistically significant. CIs are reported throughout this chapter.

Relative to the synthetic control group (Figure 2.1), there was a small decrease of -0.03 per month (95% CI [-0.29, 0.25]) in hospital admission rates (per 1 million people) for under 25s in SV-funded PFAs, but this was not statistically significant. The wide range of CIs indicates high levels of variability in hospital admissions between areas and over time. Recognising the preventative/early intervention focus of VRUs, it can be argued that impacts on hospital admissions (a key measure of SV) may take longer to materialise (i.e. there is a lag effect). Evidence from the Scottish VRU (established in 2005) supports this argument, where homicides and SV were approximately halved, but this was over an eight-year period ^{[footnote 6]}.

Figure 2.1 presents the results from the SCM for the rates of hospital admissions resulting from intentional injury caused by a sharp object for the under-25 age group (monthly PFA-level data). The SCM produced a good counterfactual with the synthetic control group (the dashed blue line) closely tracking the pre-intervention trend of the treatment group (solid black line). The treatment group in Figure 2.1 is the average hospital admission rate for SV-funded PFAs. The dashed vertical lines (left to right) represent SV funding allocated and Surge activity commencing (April 2019), VRUs set up (September 2019), and the start of COVID-19 restrictions (April 2020). For context, the increase in admission rates in June 2020 (month 101) was following the lifting of the national COVID-19 lockdown.

Figure 2.1: Average hospital admissions (under 25s) rates (per 1 million people) for treated PFAs and synthetic control group

SCMs adjusting for COVID-19 (using Google Movement data) and adjusting for Surge activity (using Stop and Search rates) also did not provide statistically significant results (and estimates were within the 95% CIs of the unadjusted SCM (Figure 2.1)).

Analysis conducted on the same hospital admissions outcome, but for all ages (not limited to under 25s), estimated an increase of 0.25 (95% CI [-0.27, 0.94]) in SV-funded PFAs relative to the synthetic control group constructed; however, this was not statistically significant.

2.2 Impact on homicides

A statistically significant impact on homicides rates was not detected. The SCM estimated an increase of 0.32 (95% CI [-0.96, 1.61]). As per hospital admissions, the CIs reported are wide, and reductions in homicides are arguably a longer-term outcome for VRUs.

Figure 2.2 presents the results from the SCM for the homicides (quarterly PFA-level data), which can be interpreted similarly to Figure 2.1.

Figure 2.2: Average homicide rates for treated PFAs and synthetic control group

2.3 Impact on police recorded violence

There was a statistically significant decrease in police recorded violence in SV-funded areas. Impacts were most prevalent for violence without injury offences, which aligns to the more preventative/early intervention nature of VRUs.

Figure 2.3 shows the results from the SCM for the violence offences and possession of weapons offence rates (per 100,000) (monthly Community Safety Partnership (CSP)-level data). The key findings are:

• relative to the synthetic control group, violence with injury rates in treated areas showed signs of reductions, but this was only statistically significant in some periods
• for violence without injury, there was a statistically significant reduction after allocation of SV funding; this impact was most prominent during live activity of both Surge and VRU
• there was a slight reduction in possession of weapons offences for treated areas, but this was not statistically significant
• the SCMs did not change substantially when adjusted for COVID-19 (using Google Movement data), indicating movement in SV-funded and comparator areas was similarly affected by COVID-19

Figure 2.3: Average police recorded offending rates for treated areas and synthetic control group)

Figure 2.4 depicts the cumulative impact from the violence offences SCMs (above), which is the total difference in police recorded offences between the treated areas and synthetic control group ^{[footnote 7]}. In other words, for a given number of months after SV funding was allocated (April 2019), the figure shows how many offences in total had been avoided (the blue line). The grey shaded areas are 95% CIs. Key findings are:

• an impact was evident for violence with injury for brief periods (summer 2019 and early 2020) but not sustained throughout 2020
• there was a reduction of 80 violence without injury offences (per 100,000 population) by September 2020 (the latest period available) - based on the total population in funded PFAs (34,321,607), it is estimated that 27,585 violence without injury police recorded offences were avoided by September 2020

Figure 2.4: Cumulative impact on police recorded violence offences, from April 2019 (start of SV funding) to September 2020

To further explore and validate the impacts from the SCM, interrupted time series (ITS) analysis was conducted. For violence with injury (Figure 2.5), there was an increasing trend in funded areas prior to introducing SV funding (left-hand side of the vertical dashed line). Following funding, the slope of the trend changed towards a slight decrease. This change in slope was statistically significant. For violence without injury offences (Figure 2.6), there was a levelling off in the trend following funding (not statistically significant). It is important to remember the ITS does not include a comparator group – the counterfactual is based on the pre-SV funding trend (dashed red line) – which explains the differences to the SCM results. The key finding is that both the ITS and SCM indicate reductions in violence offences, relative to their respective counterfactuals, following the introduction of SV funding.

Figure 2.5: ITS of police recorded violence with injury (all SV-funded areas)

Figure 2.6: ITS of police recorded violence without injury (all SV funded areas)

2.3.1 Analysis adjusting for Surge

To isolate, as far as possible, the impact of VRUs from Surge activity, SCM models adjusting for Stop and Search rates (a proxy for police activity) were employed. This means impacts of SV funding, minus that associated with increased police activity (i.e. Surge), can be estimated. However, note that Stop and Search rates only capture an element of Surge activity (it works in multiple ways to reduce violence) and data for some PFAs were missing ^{[footnote 8]}. Recognising some PFAs were removed from the analysis, the (unadjusted) impact estimates presented below differ to those in the previous section.

Figure 2.7 shows the cumulative reductions in violence offences from SCMs with (black line) and without (light blue line) adjusting for Stop and Search rates. For violence with injury offences, the impact after adjusting for Stop and Search is much closer to zero (the dashed horizontal line) than without adjustment. This suggests that Surge activity is contributing to reductions in violence with injury. For violence without injury offences, there was also a reduction when adjusting for Stop and Search rates. This aligns with the preventative/early intervention nature of VRUs, where impacts on more serious forms of violence will likely take longer to materialise. The key finding at this stage is that VRU and Surge activity appear to be complementing each other, and adjusting for Stop and Search rates can help isolate the impact of each.

Figure 2.7: Cumulative impact on police recorded violence offences (with and without adjustment for Stop and Search rates), from April 2019 (start of SV funding) to September 2020

To further understand the impact of VRUs independent of Surge activity, ITS was conducted with the intervention period starting in September 2019 when VRUs became operational (rather than April 2019 in the previous section). The results showed a change in slope (levelling off) following intervention, but this was not as pronounced as the ITS covering the entire SV funding period. Again, this supports the dual approach of prevention (VRUs) and enforcement (Surge). The additional ITS outputs are provided in Annex 3.

2.3.2 Analysis focusing on violence hotspots

As highlighted in the programme-level ToC (see Figure 1.1), VRUs (and indeed Surge) are using data and intelligence to identify violence hotspots to target resources. To explore the potential impact of this approach, SCM was applied to Lower Layer Super Output Areas (LSOAs) ^{[footnote 9]}. LSOAs are small geographical areas that have an average population of 1,500, often considered as ‘neighbourhoods’. For PFAs with the required data, analysis was undertaken on LSOAs with historically high levels (top 10%) of violence offences in SV-funded areas ^{[footnote 10]}.

The results from the LSOA-level SCMs support those at the whole VRU level (see previous sections), and indicate VRU and Surge activity are effectively targeting violence hotspots, with statistically significant decreases in both violence with (-2.8%) and without injury (-6.9%). Figure 2.8 presents the results from the LSOA-level SCM (‘high-violence’ neighbourhoods). The SCM produced a very closely matched synthetic control group. Shortly after introducing SV funding in April 2019 (dashed vertical line), impacts become visible ^{[footnote 11]}.

Figure 2.8: Police recorded violence offences for high-violence treated areas (LSOAs) and synthetic control group.

2.3.3 Cost-benefit analysis

Cost-benefit analysis was conducted, building on the results of the impact evaluation.

Over the 18-months from when SV funding was deployed and up to the latest point where offence data were available (April 2019 to September 2020), 27,585 police recorded violence without injury offences had been avoided in treated areas, relative to the synthetic control group (see Figure 2.4). After accounting for under reporting/recording in police data (by applying a multiplier of 1.5) ^{[footnote 12]}, an estimated 41,377 violence without injury offences were avoided due to SV funding.

The cost of violence without injury is estimated to be £6,480 per offence. This estimate (and that for violence with injury) is the total economic and social costs, which include the anticipation of, consequences of and response to crime ^{[footnote 13]}. Based on the number of offences avoided (41,377), the total costs avoided (benefit) were £268,125,706.

Following the same approach for violence with injury offences, an estimated 2,937 offences were avoided. After accounting for under reporting/recording in police data (by applying a multiplier of 2.6), an estimated 7,636 offences had been avoided due to SV funding. Based on the estimated cost per offence of £15,353, total costs avoided were £117,237,412.

The total costs associated with VRU and Surge activity over the same 18-month period was £121,952,973 ^{[footnote 14]}. The benefit-cost ratio (£385,363,118 total benefits divided by £121,952,973 costs) is 3.16. So, for every £1 of SV funding, there was a return on investment (as offences avoided) of £3.16.

It is important to note that the benefits are calculated based on impact estimates that have a degree of uncertainty. This uncertainty is captured within the CIs reported for each estimate (see Figure 2.4). For completeness, benefit-cost ratios were calculated for the upper and lower CIs. The benefit-cost ratio based on the lower CI was 11.07, and 0.4 for the upper CI; so, considering the uncertainty in the impact estimates, the true return on investment likely falls within the range of £0.40 to £11.07 for every £1 invested.

2.4 Wider evidence of impacts on serious violence

In this section, wider qualitative evidence from consultations with VRU stakeholders on the impact of VRUs on SV, the interaction between Surge and VRUs, and the impact of COVID-19 are considered.

2.4.1 VRU Perceptions of the impact on serious violence

Overall, VRU stakeholders were generally hesitant to attribute any changes in SV trends to VRU activity in year 2. This was due to several factors including perceptions of the impact of COVID-19 on SV incidents; the preventative nature of VRUs (relative to the enforcement focus of Surge); and VRU stakeholders feeling that the first two years of activity were having more of a tangible impact on local structures and co-ordination around the SV agenda. VRU stakeholders often felt that it was too soon to say if and how this had translated to changes in numbers, even if they were experiencing reductions in SV incidents locally.

“We can evidence that certainly knife crime is levelling out in [VRU area] and not continuing on its upward trajectory. I don’t think we can pinpoint any one bit of work though, that would say that’s causing that. It’s a whole package, and I guess it’s always going to be quite difficult with a whole-system approach to it.” Stakeholder

Despite challenges in speaking about the contribution of VRUs at the area level, in some VRUs, stakeholders noted they could link specific interventions to a decline in incidents. For example, one VRU tracked police interventions at the six- and 12-week markers of a programme for young people at most risk. The evidence showed a decline in the involvement of anti-social behaviour and criminal activity at the time points. The scale at which interventions were operating also affected VRU stakeholders’ perceptions of impact. For example, one VRU frontline stakeholder working in a village noted they had seen demonstrable changes in a young person’s life outcomes as a result of their project but, given the scale, it would not result in change at the VRU area level.

As discussed further in Chapter 4 (MA Working), in year 2, many VRUs were doing a lot of work to embed a trauma-informed approach across agencies’ workforce, including focusing on the impact of Adverse Childhood Experiences (ACEs). Several stakeholders felt that this work would lead to system-wide changes in social work and education but noted that this would take years to translate into reductions in SV.

2.4.2 VRU and Surge interaction

As discussed further in Chapter 4 (Multi-agency working), in year 2, VRUs and Surge have increased their alignment at both the strategic and operational level, co-ordinating approaches and activity where possible. For many stakeholders, this increased alignment made it difficult for them to disentangle the impact of VRUs versus the impact of Surge on levels of SV. Where stakeholders could disentangle the two, they pointed towards Surge funding facilitating reductions in SV incidents over a short time. They found it easier to highlight the link between a particular Surge operation and subsequent changes in incidents. For example, in one VRU, an operational stakeholder highlighted that Surge funding had made a demonstrable impact on the reduction of crime locally, as they could use the funding to pay for police officer overtime so there was greater police presence on the streets.

“Actually having this [Surge] funding and allowing us to put extra cops on the street, especially during this time, has proved invaluable for us.” Operational stakeholder

Despite Surge funding having an impact in the short term, it was recognised that without being sustained, activity would increase again eventually:

“If you flood an area with police and other crime prevention functions you will have a quick impact. It doesn’t last, but at least it keeps people safe for that period.” Strategic Stakeholder

However, where stakeholders felt Surge funding activity was having an impact, they also emphasised the important role that VRU activity had played, such as identifying crime hotspots or developing awareness-raising campaigns about particular crimes alongside Surge-funded activity on the ground.

2.5 Conclusions

There is evidence of emerging impacts of SV funding, particularly for violence without injury offences. This is encouraging for VRUs (and supported by qualitative evidence) where the focus is on early intervention/prevention. Regarding SV (i.e. hospital admissions and homicides), statistically significant impacts were not detected. It will be important to monitor impacts on SV in the future, when it is anticipated the more preventative approach of VRUs will affect these longer-term outcomes. It should also be noted that the rates of SV do not appear to have increased in funded areas (relative to non-funded areas), which is a positive finding. The dual approach of prevention and enforcement from VRUs and Surge, respectively, appears to be effective and complementary.

3. VRU design, focus and models of working

This chapter describes the approach taken by VRUs to define SV and assesses the progress made by VRUs in identifying at-risk cohorts. It also sets out the evolution of the VRU models of working over year 2. The resultant effectiveness and contribution/consequences of these foci and models are discussed in the subsequent chapters of the report.

3.1 VRU definitions of serious violence and the at-risk cohorts

VRUs form a key component of the Serious Violence Strategy (2018) and, as a result, their priorities and associated definitions of what constitutes SV reflect those set out in the Strategy, associated Home Office guidance and the VRU key outcome measures (see Chapter 1 for more details). This formed the basis of the types of evidence that were drawn together over year 1 – including both hard and soft data from a range of sources – to develop SNAs. The resultant SNAs varied considerably in depth and quality, but as a minimum, sought to begin identifying the primary at-risk cohorts (i.e. individuals or groups of individuals that were at risk of becoming or remaining involved in SV) that were prevalent in each area and the hotspots (and hot times) where incidents were most likely to take place.

The SNAs provided an initial evidence-based foundation upon which to develop the first set of Response Strategies, that in turn informed the focus and activities for year 2 of the programme.

The research conducted in year 2 found that the SNAs and Response Strategies provided a deeper understanding of at-risk cohorts, an overview of existing services and a better understanding of the gaps in provision. This included some validation of the perceived key drivers of SV identified in the year 1 process evaluation, which most commonly included ACEs, deprivation/poverty and austerity measures which impacted the level of resource available to prevent and/or reduce SV.

Key findings and emerging evidence from the SNAs and Response Strategies were circulated to a variety of stakeholders and partners across most VRUs to raise awareness of the purpose and focus of their activities, and to foster engagement amongst a wide range of organisations. The reach of this specific communication varied considerably across the VRUs, with some promoting the findings amongst statutory partners, wider community and voluntary organisations and frontline practitioners that worked with the identified at-risk cohorts. Conversely, other VRUs had taken a more conservative approach which had, in the main, cascaded findings across the partners represented on their VRU Governance Boards and on the core VRU teams. This had led to lower levels of awareness of the detailed intentions of the relevant VRUs amongst some wider stakeholders and frontline practitioners, which is likely to have limited the scale of collaboration that took place within these areas.

The communication activities often involved workshops which included information sharing and feedback sessions, to engender a sense that the SNAs and Response Strategies were ‘live’ documents that would be updated on the basis of evolving local contexts, a growing evidence base and softer intelligence gathered from partners.

“We do sometimes have a conversation about do we need to revisit the serious violence definition? So, night time economy. We know community safety partnerships, have plans around that. We know there’s work around the town. High streets. Some of the funding that’s going in through the local economic partnership. So, it was like where’s the gap in provision? A lot of people who sit around our strategic board obviously sit around those other strategic boards as well [and can help identify the gaps], so this is where we need to go. The SNA has assisted us with understanding that a bit more, but it is something we need to keep under review.” VRU lead

Taken in the round, the newly established SNAs and Response Strategies and the associated communication activities resulted in many VRUs reviewing and refining their SV definitions at the outset of year 2, to:

• better reflect local need
• avoid duplication where established networks and provision were working well
• add value by addressing gaps in provision
• expand existing services where required

Consequently, some VRUs broadened their SV definition to address high incidences of specific types of SV, including those resulting from the night-time economy, county lines and domestic violence. Conversely, others narrowed their definition to avoid unnecessary duplication; for example, several VRU areas had sufficient existing domestic violence networks which freed up VRU resources to tackle other areas with little or no coordinated provision.

“There’s quite a strong structure put in place for domestic violence based on the work done 15-20 years ago. There was almost a view of domestic abuse is incredibly important, but it has its own governance framework and strategic view. So, we do not want to repeat that. We want to be a complementary piece. More generally, it was a view of like what are other parts of the system doing and where’s the gap. The gap tended to be that community violence out of the household and what did it look like? And that picture around county lines and gangs which is so difficult to measure.” VRU lead

In addition, although VRUs continued to focus their efforts on supporting young people up to the age of 25, the majority were not exclusively supporting this group, so they could offer some support to older beneficiaries where relevant. This included an expansion of work with vulnerable families in recognition that holistic approaches to supporting those most at risk of becoming involved in SV should involve whole family units. It also included a renewed focus on repeat offenders (both young people and adults) that were often ‘getting lost’ in the transition between youth offending services, the prisons estate, probation services and adult services.

“Something that we’ve got data on is around repeat offenders and repeat victims, and that’s something that we have seen a trend increasing. I think that’s quite important because actually, that suggests that those people, their risk factors for violence haven’t been addressed so therefore, they’re continuing to be violent, whether that’s related to organised crime or something else. Obviously, it’s a smallish proportion of the total of either victims or offenders, but it’s an increasing proportion and we’ve seen that over the last year, actually, that that’s increasing.” VRU Core Team Member

The SNAs also enabled VRUs to adopt place-based and people-based approaches to enable more effective targeting of their resources. These approaches were set out to varying degrees in VRU Response Strategies that were designed to articulate how they intended to respond to the needs outlined in their SNAs.

• place-based approaches – target an entire community and aim to address issues that exist at the sub-area (place) level (e.g. specific neighbourhoods); for example, some VRUs were targeting areas with high levels of deprivation, low attainment, high school exclusion rates and high levels of crime, with an aim to intervene at the earliest possible moment using a holistic approach to preventing SV
• people-based approaches – target specific groups of people or characteristics of a group of people; for example, targeting those already known to be involved in SV, or identifying at-risk cohorts such as young people on the verge of becoming not in employment, education or training (NEET) or those with behavioural problems or have experienced ACEs

Impact of COVID-19

Year 2 of the VRU programme has been significantly impacted by the COVID-19 pandemic and associated national and local lockdowns and restrictions. Nearly all interviewees consulted as part of the evaluation discussed the challenges the pandemic had posed, which at the broadest level included having to navigate disruptions and changing trends in the prevalence of different types of crime and SV. This included significant reductions in public violence over some months, alongside rising rates of what were commonly described as ‘less visible crime’ such as domestic violence and drug related-crime (see Chapter 2).

“What I think covid has potentially identified more so, is the violence in the home, domestic violence, violence against women and girls (VAWG), trends showing through our specialist charity networks an increase in calls and increasing requests for help and intervention in that space. I don’t think its necessarily changed the data… but its definitely changed the emphasis that brought it to the fore in just a very practical way – that street violence through covid and lockdown obviously declined and that exposed the fact that a lot of violence takes place in the home environment.” VRU Core Team Member

Many VRUs therefore adapted some of their work to address these new and somewhat unprecedented challenges, which involved a re-prioritisation of how some of their resource was deployed (see Chapters 4, 5, 6 and 7 for more details). Looking forwards, several interviewees from across the VRUs voiced their growing concern about the long-term effects of the pandemic, such as the economic downturn, rising levels of youth unemployment and the detrimental impacts this would likely have on mental health. They reflected that this would pose additional challenges for the work of the VRUs in the future, which may be increasingly reflected in the evolving nature of their at-risk cohorts moving forwards.

“What we have now is really worrying situation – what COVID has essentially done – we don’t know actually what is going to be created in terms of society after all of this has gone – what we do know is the economic downturn, the mental health, alcohol use, god knows what we are going to face.” VRU Core Team Member

“You’ve now got the double whammy really with the after effects of COVID, in terms of this work was hard enough in times when we were coming out of austerity. It’s going to get much harder in times when it’s getting worse and demonstrating to a vulnerable young person who’s, ‘Oh, do I go to school today and crack on with my education and get my exams, and all the rest of it, or do I go over there with that bloke who’s offering me the trainers and telling me that I can go running for him and I can be taking home £100 this evening?’ In the context we’re now going into it arguably becomes harder to influence that decision.” VRU Core Team Member

3.2 VRU models of working

The year 1 process evaluation report highlighted that most VRUs comprised at least three ‘layers’, which included a governance board, a dedicated team of strategic/operational staff commonly seen as ‘the VRU’, and an additional more localised interface that operationalised activity on the ground. For most VRUs, this manifested in a single/centralised unit model, though a small number had a hub-and-spoke structure where a core VRU team sat at the centre of local VRU teams responsible for local delivery. Regardless of the VRU structure, all VRUs were working closely with existing partnerships and local organisations to develop a more prevention-focused, holistic and coordinated approach to addressing serious youth violence.

In year 2, the structures and models utilised by VRUs were largely unchanged from year 1. One exception to this pattern was noted: where two VRUs were moving from a single/centralised structure to a ‘quasi-hub-and-spoke’ approach to enable the development and delivery of more place-based approaches. This transition was viewed more as an evolution in their model of working as their delivery has matured, as opposed to a result of their previous approach being ineffective.

A second year of delivery had also provided interviewees with more time to reflect on the challenges and benefits arising from their adopted approaches. For example, for hub-and-spoke VRUs, this included recognition of the need to ensure there was genuine co-ordination from the centre with common themes applied across the spokes, even where there was some variation within this in terms of local focus.

“There are elements that need to be pan-[region] and we see the value in that, there is a pan-[region] police force so this makes sense. But we also we have three distinct authorities that have their own needs. It is helpful to have that pan-[region] discussion but then be able to move to our local discussions and have that local steering moving forwards.” VRU lead

The balance between creating a regional approach while still allowing local areas to retain autonomy over their own needs and priorities was also a consideration in VRUs with centralised models; interviewees saw this as important to encourage stakeholder buy-in. For example, to address this, one centralised VRU had developed an elected members’ reference group, with representation from all component local authorities. This approach ensured local needs and priorities were reflected in VRU activities.

Strategic governance

In terms of strategic-level governance, the research reinforced the importance of executive boards and steering groups for facilitating true multi-agency working. Generally – as one might expect from a whole-system approach – these boards had broad, usually senior-level representation from partner agencies, including those from criminal justice, youth justice, education, health and safeguarding.

The main changes to this layer of the VRU infrastructure in year 2 included:

• a broadening of both the number and type of agencies involved, to support the development of collaborative delivery models that prevent duplication and complement existing service provision; the research found some examples of representation of service users and local communities at strategic boards; for example, one VRU had a board member with lived experience, and another had local-level steering groups working directly with local communities; a third had developed a Violence Reduction Citizens’ Advisory Panel, which feeds directly into the strategic board; some VRUs had also sought to ensure that other agencies, such as Voluntary Community and Social Enterprises (VCSEs) and faith groups, were represented at this level, and there also appeared to be increasing representation from the Department for Work and Pensions (DWP) in a minority of cases
• the streamlining of wider governance boards to reduce duplication and enable a greater pooling of strategic resource to drive the relevant agendas; for example, one VRU streamlined an existing prevention board into their VRU board to ensure a more effective use of their resource
• improvements in the clarity of remit and responsibilities of the governance boards, reflecting a growth in maturity of VRUs and the associated embedding of whole-system approaches to tackle SV

No clear differences were identified between the governance structures associated with the different models of working.

Operational structures

At an operational level, the research found that core VRU teams had also expanded in terms of size and composition as year 2 progressed, with recruitment activities particularly significant in the first half of the year. Generally, and as seen in year 1, the teams had a lead, an analyst or team of analysts and researchers, public health colleagues and operational leads overseeing different aspects of activity, as well as support staff. Expansions to the core teams varied considerably across the VRUs. This included the recruitment of staff from the local Integrated Care System, children’s social care, a domestic abuse specialist, local evaluators and greater representation from the VCSE sector.

As per the findings in year 1, teams were often co-located (that is, prior to COVID-19 restrictions being implemented), and many staff had been seconded in from partner agencies, either on a full- or part-time basis. In the majority of such cases, interviewees noted that their ‘home’ roles (i.e. original roles within partner agencies) were complementary to their role in the VRU and the move had been beneficial.

The most significant change across many of the core VRU teams involved restructuring around the key priority themes or workstreams that had been identified through SNAs and Response Strategies. This form of restructuring was prevalent across both single/centralised and hub-and-spoke models, and had most commonly involved the appointment of an operational lead for each theme/workstream, who reported to the VRU lead and/or into the strategic boards via a ‘workstream sponsor’ at board level. The number and type of themes/workstreams varied across the relevant VRUs, though generally could be categorised as either operational- or policy-focused. Typical workstreams included data and evidence (including evaluation), developing capacity in communities and workforces, trauma-informed and early intervention approaches, approaches focused on violence hotspots and enforcement activities, and communication and campaigns.

Staffing challenges

Staff turnover remained a common issue in year 2, which was exacerbated by COVID-19 where staff (particularly those from public health and policing) were redeployed to their home teams to support the national response. In addition, sustainability of teams comprising primarily seconded staff also remained a concern, particularly as teams approached the point at which secondments were due to expire.

“A key organisational change has been that our public health analyst has been redeployed to the local authority due to COVID-19. Although this is unlikely to have a large impact on us, the SNA refresh and other parts of the research and evaluation strand are likely to be delayed and pushed back beyond their original schedule.” VRU lead

The root cause of both issues (when discounting for the impact of COVID-19) was perceived to be related to the short-term nature of the funding for VRUs, which led to instability and insecurity of staffing.

“This is a classic example of when you have cliff edge short-term funding, and you have good people in your team, people will always be looking over their shoulders for other things, because especially in the current climate, people need security.” VRU lead

Impact of COVID-19

While COVID-19 impacted negatively on staffing, there were some positive impacts of the pandemic in terms of providing VRUs with some ‘breathing space’ to develop their models and ways of working in more detail. Interviewees noted the pandemic had provided a ‘natural pause’ to reflect on their approaches, which had proved useful given the relatively tight timescales for the programme. Core VRU teams referenced close monitoring of the impact of COVID-19 on delivery via risk registers.

4. Multi-agency working

This chapter explores VRUs’ approaches to multi-agency working, including stakeholders’ views on the effectiveness of these approaches. It also examines the contribution of multi-agency working to VRUs’ overarching aims, both from a strategic and operational perspective.

4.1 Approaches to multi-agency working

All VRUs have focused on establishing and/or developing partnerships with a wide range of agencies. This section discusses how VRUs have worked with these partners both strategically (in terms of leadership and co-ordination) and operationally (through devolving funding, providing non-financial support and supporting professional development), and the progress made since year 1.

4.1.1 Strategic multi-agency working

VRUs typically had strategies underpinning their approach to multi-agency working. However, evidence-based principles shaped approaches in only a few cases. In two cases, VRUs used academic research and Public Health England’s ‘5Cs’ to frame their approach ^{[footnote 15]}. Other VRUs had already engaged with the Scottish VRU to identify how they could implement a similar model in their local area.

All VRUs took steps to develop shared strategic aims across partners, though approaches varied. Engagement included consultation – for example through workshops, co-production in developing SNAs and Response Strategies – and liaison to develop definitions of SV and the at-risk cohort. In all cases, board meetings were a key mechanism to maintain strategic engagement from partners. Police representatives were often on VRU boards to support the strategic alignment between VRU activity and Surge activity. Some VRU leads also attended Surge board meetings to further identify synergies and opportunities for collaboration. Regular meetings between VRU and Surge leads allowed discussions about each programme’s priorities and how to align them. In one case, the Surge lead was co-located in the VRU team and could align Surge spending with VRU activity (e.g. by allocating additional policing resource in hotspots that VRU analysis had identified). While this alignment was present in some VRUs in year 1, stakeholders noted how it had strengthened over year 2.

Similarly, approaches to multi-agency working developed in year 1, both building on existing partnerships and developing new ones, continued to develop and were strengthened in year 2. More work was done in year 2 to map existing structures and partner resources. This helped to identify and reduce the extent of duplication and siloed working across agencies. Linked to this, giving partners responsibility for workstreams helped ensure accountability and support for strategic VRU aims, whether workstreams were first developed in year 1 or new to year 2. For those building on year 1, more work had been done to operationalise the workstreams, for example by creating clear leadership and governance structures or focused operational groups. VRUs also facilitated new partnerships between agencies, in one case linking health partners with the police to develop an approach to improve patrolling in SV hotspots.

For most VRUs, the approach to multi-agency working sought to engage and involve partners across different localities. Including voluntary sector and/or community representation on boards was also acknowledged as important, whether informing VRUs’ strategic vision or to sense-check that strategy worked operationally. The ‘hub-and-spoke’ models of some VRUs naturally facilitated a locality-based approach to multi-agency working. In all of these cases, the hubs set the overall strategic direction of the VRU and the spokes operationalised this within their local context. Equally, there was evidence of multi-agency working across localities in more centralised VRU models. Several devolved the responsibility for developing and implementing local violence reduction strategies to CSPs. Others gave core team members responsibility for developing partnerships in specific geographic areas.

4.1.2 Operational multi-agency working

As discussed in Chapter 3, many VRU core teams seconded in staff from partner agencies to promote partner buy-in, as well as ensure operational-level co-ordination. Police secondments were viewed as helping to support operational co-ordination between VRU activity and Surge provision. Co-locating stakeholders also supported effective multi-agency working.

VRUs provided professional development opportunities for core team and partner agency staff as a mechanism to support consistent ways of working. For some VRUs, this included training and development to facilitate a trauma-informed approach ^{[footnote 16]}. For example, one VRU worked with local prisons and provided trauma-informed training to staff in the custodial estate. Devolving funding to partners also formed a key mechanism to facilitate multi-agency working at the operational level. Several VRUs devolved funding to CSPs to implement local Response Strategies, while funding local community organisations to respond to SV in specific locations/neighbourhoods was also common.

4.2 Effectiveness of approaches to multi-agency working

Across all VRUs, activity to develop multi-agency approaches was generally seen as effective. Overall, VRUs appeared to have made good progress in year 2 in building on existing partnerships and aligning structures, strategies and partners. There was consensus that VRUs had effectively mapped out and consolidated meetings and partnership mechanisms. However, there were still some concerns about activity being duplicated and partners being overburdened by VRU responsibilities. Strategic and operational stakeholders with this view often spoke about challenges for partners for whom SV is not a ‘focus of their day job’, perhaps indicating that more work is needed to truly embed the whole-system approach.

There was strong evidence of VRUs making progress in encouraging partners to take shared ownership of strategic aims. This progress occurred throughout year 2 and it was common for strategic stakeholders to flag that this process took time, particularly with some partners such as health and education, who often had less engagement due to having to focus on the immediate COVID-19 response.

Across strategic and operational stakeholders, there was generally a consensus that VRUs had been effective in providing partners (and their staff) with support and resources to engage in a multi-agency approach. VRUs were likewise viewed as an important additional resource to oversee and co-ordinate activity. However, some stakeholders within hub-and-spoke or locality models felt that occasionally the model could complicate the implementation of a multi-agency approach and hence compromise effectiveness. The sheer size of the VRU area and competing local/spoke-VRU priorities were cited as factors in this.

4.2.1 Enablers of effective multi-agency working

Additional enablers for multi-agency working, beyond those detailed above, included:

• perceived credibility of the VRU in engaging partners, as well as senior representation from those partners, helping to reinforce the multi-agency, partnership-based nature of the VRUs

“Whether it’s programmes or policies, people feel like they can trust that because it’s coming from a place that’s been done, ultimately, with the backing of the Home Office… it is a partnership-led initiative, so I think that’s been really important, that leadership and communications aspect of us having a common goal to reduce violence.” Operational external partner

• although the involvement of different partners was important, the perception of the VRU being a unit independent of the police was an important enabler for building trust with communities and community organisations

• as well as creating shared ownership through initial consultation work and co-production to inform SNAs and Response Strategies as outlined above, publishing these documents indicated outwardly that partners were engaged and supported the VRU’s aims

• clearly articulating the approach helped agencies to decide whether they would engage in the partnership; this was especially important for operational and frontline stakeholders; strategic stakeholders said it was important to get partners to understand that “they are a cog in a much bigger wheel”; for partners to see where their role fits in the SV reduction agenda alongside other agencies; operational stakeholders noted that a concise SNA and response strategy, clear terms of reference setting out responsibilities, sharing information, government guidance and strategy, all helped to communicate the approach

“For me, open honesty [and] commitment to work as a partnership is really important… it’s around that sharing of information and having a rich picture of information to actually tackle the most significant issues.” Operational external partner

• the interpersonal skills, drive and leadership of VRU leads was a key enabler to keep strategic members engaged; strong relationships established prior to VRUs facilitated multi-agency working, reiterating the importance of the person leading the VRU; however, it was acknowledged that this may not be sustained should they leave

• additional focused resource from the VRU to co-ordinate activity was noted as a key enabler; VRUs helped stakeholders to ‘focus minds’ and to identify opportunities for collaboration; ultimately, this was seen as supporting the development of a whole-system approach to tackle the cross-cutting issue of SV

“The fact that there is this pure focus on serious violence is very helpful as a way to bring people together and put fresh eyes on a particular issue, because if you’re siloed into one way of working, say around crime, youth crime, you’re not necessarily looking wider. Or, you’ll know that trauma’s a factor, but when you’ve got a VRU, it’s enabling more focus on trauma.” Strategic stakeholder

• strategic stakeholders highlighted the role (and potential further role) of wider policy that facilitated multi-agency working to tackle SV; some partners from statutory agencies mentioned the Serious Violence Duty (Home Office, 2019), which required them to develop their own SNAs and thus inform their response; they also felt the VRU added value by supporting agencies to access different datasets and enrich their analysis, although there was a concern about potential duplication

• with the sustainability of the whole-system approach in mind, some VRUs have tried to ensure that they support multi-agency working at the operational level through engaging local community organisations; several VRU leads highlighted how, through commissioning interventions, they were trying to build the capacity and capability of the community sector; for example, one VRU-supported networking between grassroots organisations to facilitate the development of a consortium that could later bid for funding/work

• strategic and operational stakeholders from several VRUs commented that a positive consequence of COVID-19 and the shift to virtual working had facilitated closer multi-agency working in some ways because people could make meetings more regularly than they could before COVID

4.2.2 Barriers to effective multi-agency working

Barriers included:

• competing strategic priorities within partner organisations meant they could not always take full ownership of the VRU’s strategic vision; stakeholders noted that there was a risk, especially for statutory organisations (e.g. police, local authorities), that changes in political leadership could affect engagement

• a perceived lack of central government direction, fully embedded across all departments, was viewed as a barrier; it was felt that greater cross-departmental buy-in and direction of the violence reduction agenda would ensure a stronger mandate for all partners to be involved and to take more ownership and accountability for delivering the vision

“If… the centre of Government were seeking a multi-agency response then we should be able to get briefings about the existence of the VRU as well as leverage and encouragement to engage to come down through the NHS, the Department of Health, the Department for Education. It feels to me entirely absent at the moment, this encouragement and support for the VRUs to be multi-agency, through any government department bar the Home Office.” Strategic Stakeholder

• stakeholders in multiple VRUs still had concerns that the VRU was too police-led, which acted as a barrier to engagement from other partners; strategically, this meant that some partners took a step back from taking ownership of the response, while operationally it could erode trust; for example, one VRU employee highlighted how their email address had ‘police’ in it and felt that this could be a barrier

“I have ‘police’ in my email. If I’m asking a local authority to share details of data around children and children services, I would not be surprised if they were slightly hesitant, at least, if not completely reluctant, to share data for fear of what they might see as ‘the police’ and what we were going to do with it right now.” Operational stakeholder

• while progress had been made, frontline and operational stakeholders across multiple VRUs felt that there remained a lack of strong strategic involvement from grassroots voluntary and community sector (VCS) organisations, who they felt were close to and trusted by communities, and more able to be agile and responsive to local needs; some operational stakeholders felt that VRUs’ strategic boards did not necessarily understand the diversity of the work done by grassroots VCS organisations, and the potential benefit they could bring if involved more at the strategic level

• varying business/budget planning cycles across partners were seen as a challenge to mainstreaming long-term strategic multi-agency working in some contexts; alignment with the VRU funding period (i.e. the financial year) was perceived as being difficult and creating a barrier to long-term planning; this was felt to exacerbate the challenge around strategic priorities, as business cycles shaped how partners prioritised funding

• there was strong consensus that short-term funding was a key limitation to cultivating long-term multi-agency approaches; at the strategic level, it was felt to reduce partners’ confidence in the model and make them less inclined to invest in it (because of uncertainty about central funding); for operational staff, it affected engagement with other partners due to a lack of confidence that the VRU would be around in the long term; stakeholders from frontline organisations commonly felt that the short-term nature of funding impacted on their perception of whether it was worth their time and effort to engage

• some strategic stakeholders were concerned that the funding for interventions was spread too thinly and the amounts on offer could dissuade organisations from engaging in the whole-system approach; for example, one VRU lead noted that there was underspend in their community grants budget because VCS organisations had not applied due to small funding-pot ‘fatigue’

“When you’re dealing with such a short burst of funding, it puts off some potential projects or interventions because it’s too short. More secure funding would be helpful.” Operational stakeholder

• operational staff and strategic partners reflected on a continued cultural challenge across agencies, related to a lack of common language; one VRU lead highlighted how terms such as ‘prevention’ or ‘trauma-informed’ mean different things to different organisations or professionals and, therefore, work was needed to align partners’ understanding

• statutory agencies’ responsibilities sometimes affected relationships between partners, given that statutory partners have specific legal duties, working parameters and approaches to risk that VCS organisations do not; however, there was recognition from some that this was a challenge inherent to multi-agency working across statutory and non-statutory organisations

• stakeholders continued to have some concerns about operational partners working in silos (year 1 finding)’ reportedly, operational staff in partner agencies were at capacity, and had little time or headspace for reflection on how they could work differently; there was strong consensus that activity was being duplicated and not enough had been done to link up delivery providers to work collaboratively

• not having the ‘right’ people involved in strategic discussions was seen as a key barrier to operationalising multi-agency working; some strategic stakeholders reflected that they required people with sufficient seniority to instigate change within their organisations, but also close enough to the frontline to know the best approach to implement it

“[With multi-agency working] if you get people who are too senior then what happens is that they don’t really know what they’re talking about but they’ve got lots of authority. Whereas, if you get people that are too junior, they know exactly what needs to be done but they haven’t got the authority to actually deliver their agency. So, we need to make sure that we get the right people represented from the different agencies.” Strategic stakeholder

• although remote working because of COVID-19 enabled more consistent attendance at meetings, stakeholders generally felt it was a barrier to effective multi-agency working; at the strategic level, partners were often consumed by the COVID-19 response and were less able to engage; at the operational level, it prevented co-location and being able to come together to celebrate progress and raise morale

4.3 Contribution of multi-agency working to VRU aims

An overarching aim / intended impact of VRUs is to establish a sustained whole-system approach, through effective multi-agency working, and organisational or cultural change towards the SV response. Overall, in year 2, data from stakeholders at all levels indicates VRUs have generally made substantial progress in this. There was some evidence of cultural change occurring across partner agencies, due to VRUs providing training, aligning practices and developing resources to guide partners’ work.

Across the 18 VRUs, most stakeholders at all levels (i.e. strategic, operational and frontline) agreed on the VRUs’ contribution to strategic and operational multi-agency aims. Generally, stakeholders agreed that VRU activity had made a greater contribution at the strategic level, likely reflecting VRUs being at an early stage in the long-term systems change approach. There were substantial differences in VRUs’ ‘starting points’ with multi-agency working. At the point of receiving Home Office funding, some VRUs had already established or were developing their approach and therefore already had partnerships in place. Others had limited history of multi-agency working before the VRU programme. With these contextual factors in mind, this section discusses the extent to which multi-agency working contributed to VRUs’ aims.

4.3.1 Contribution of multi-agency working to the whole-system approach

There was strong evidence (from interviews with stakeholders at all levels) to suggest that multi-agency working has helped partners and professionals to buy into and support the whole-system approach, particularly at the strategic level. In almost all cases, stakeholders reported that the VRU had added value by improving strategic links between statutory agencies, either by focusing existing partnerships on the SV agenda, or by identifying opportunities for new partnerships. It was common for stakeholders to reflect that some of the work developing a whole-system approach would have happened without the VRU, but not to the same extent and it would have taken much longer. Partners felt that they would not have had enough capacity to lead and develop this agenda beyond their own focused work; the VRU was thus seen as essential in adding capacity and resource to galvanise multi-agency working.

Some stakeholders flagged that while partners bought into the ethos of the whole-system approach, they had not yet committed to it financially, by providing staff time or intervention funding, although many planned to do so. For example, one VRU lead noted their plans to increase consultation with education, probation and early-years teams to align their work and budgets more closely in year 3. Another lead planned to work with local authority CEOs to explore funding options. Arguably, the uncertainty of future financing poses a threat to how far the SV agenda can be embedded in business as usual for partners.

From an operational perspective, there was less consistency in stakeholders’ views about how far multi-agency working, facilitated through the VRU, contributed to embedding a whole-system approach. Interviews with frontline and operational stakeholders across many VRUs suggested that staff had improved understandings of their roles in relation to reducing SV. There was also indication of some improved links and day-to-day working practices due to VRUs’ co-ordination efforts. However, operational or frontline stakeholders often also highlighted a perceived ‘disconnect’ between the VRU’s strategic vision and what was happening on the ground, in terms of the whole-system approach fully flowing through to, and informing, their day-to-day work.

Another intended outcome of VRUs is to support partner and community organisations to have increased capacity, efficiency and effectiveness. There was evidence in some VRUs to suggest this was happening, but overall, this appears to be an area that VRUs need and intend to develop more. Through streamlining and co-ordinating activity, VRUs were generally supporting partner organisations’ efficiency, ensuring more focused discussions and reduced duplication. However, capacity within partner organisations remained an issue across multiple VRUs, emphasising how VRU teams can provide additional resource helped with engaging partners.

4.3.2 Cultural change occurring across partner agencies

Across most VRUs, it was too early for stakeholders to suggest that there had been complete cultural change across partner agencies. As noted above, shared understandings of concepts and language remain a barrier. However, there was some evidence that VRUs had supported partners/professionals to better understand and respond to local needs. Equally, VRUs were working hard to develop common understandings of preventative approaches to SV across the workforce. Stakeholders in multiple VRUs spoke about how VRU-led workforce development support catalysed partners (especially schools and policing) to reflect more on how they can embed preventative approaches.

“There has been an acceptance that even from within the police – and our colleague in the [police] that works in that prevention space, would say ‘we still have a long way to go but there is a cohort emerging from the police now who get the prevention agenda and understand it and that it is relevant and important’ – and that is a massive impact really.” Operational stakeholder

4.3.3 Multi-agency contribution summary

Overall, VRUs have contributed strongly to supporting effective multi-agency working at the strategic level. They have done this by building on the work in year 1 to strengthen partnerships, streamline approaches and identify and facilitate areas for collaboration. Engagement of less involved VCS organisations (particularly at the grassroots level) and statutory partners (e.g. health, education and housing) will likely be a focus for VRUs in year 3. Given the focus on strategic alignment in year 2, there appears to be greater variation across VRUs in terms of the contribution of multi-agency working at the operational level. However, there has been some good progress in relation to developing partners’ shared understanding and skills around preventative approaches. As VRUs have strengthened their strategic alignment with partners, greater progress in operational multi-agency working could be anticipated in year 3.

“It’s been great, because across [the VRU area], there’s now a real strategic and operational consensus that this is the right thing to do, that it needs investment, and it needs prioritising across the system… The strategic leadership is fully behind a child-first approach to this criminal justice issue, which then operationally means that sort of response is much easier. It’s much less difficult to operationalise and make happen on the ground with people that are actually doing the work.” External Operational stakeholder 

5. Data and intelligence sharing and analysis

This chapter explores approaches to data sharing and analysis among VRUs and partners, the extent to which data are being used effectively, and assesses the contribution of data and intelligence sharing to meeting VRU aims.

5.1 Approaches to data sharing and analysis

VRUs have shown different approaches towards data and intelligence sharing since year 1, with evidence of continued progress in year 2. Since the first year of funding, the aim has been for VRUs to access and use multiple data sources and intelligence to inform their SNAs, develop new data-sharing agreements (DSAs), and ultimately make evidence-based decisions in terms of the VRU’s strategy to reduce violence, in particular, intervention commissioning.

5.1.1 Data sources and types

Police data are the key data source for all VRUs, largely owing to the relative ease of access (via police links). This data continued to be a key source for developing SNAs in year 2, building on such use in year 1. Increased partnership working with Surge, facilitating wider intelligence sharing (e.g. on emerging hot spots / times) has also been a key development. This wider intelligence sharing, alongside a more targeted approach to the data and its use, reflects strengthened multi-agency working and a refined strategic focus. For example, some VRUs mentioned an enhanced focus on analysing data for specific projects/interventions – “a focus on ‘what can we learn’ from the data, rather than focusing on gaining as much data as possible”.

Some VRUs also reported good progress in accessing health data (e.g. hospital admissions, ambulance data, accident & emergency), which was a challenge in year 1, with this often requiring close working with NHS colleagues to overcome barriers (see section 5.2). VRUs often recognised accessing and making use of health data as a strategic priority. There is some evidence of progress in the sharing of other data sources (e.g. education, youth justice), reported to be more limited in year 1. In particular, following production of the SNAs, VRUs reported increased understanding of what the available data cover and where the gaps are. This has enabled a more focused approach to data sharing.

The shared and accessed data are mainly at an aggregate level, enabling an understanding of hot spot areas and the at-risk cohorts. A few VRUs had began accessing individual-level data from partners, hence supporting additional analysis.

5.1.2 Data use and analysis

VRUs are accessing a wide range of data sources and monitoring multiple indicators/outcomes from these. Furthermore, the importance of more qualitative intelligence (e.g. from Surge teams and community organisations) is recognised.

“I think sitting on the [Surge board meeting] is very important for someone from the VRU… we hear the bigger picture, hotspots, gang-mapping, rating of gangs, activities, emerging new issues. That’s quite important for what we need.” VRU stakeholder

VRUs are using data to develop multiple outputs, which help inform their strategy and commissioning of interventions (i.e. Response Strategies). These include:

• the revised SNAs
• regular monitoring/performance reports/dashboards
• bespoke analysis to understand more about / identify those at risk

As well as helping develop strategies and interventions in general, these outputs have helped VRUs to understand increases and changes in the vulnerabilities being faced by the at-risk cohorts resulting from COVID-19.

To ensure the VRUs can respond to evolving needs and context (e.g. emerging hotspots), more regular outputs have been developed. Most VRUs reported sharing analysis in multiple forms, the most common being quarterly or monthly monitoring reports, and interactive dashboards. While these outputs are, owing to data availability, typically police and health data focused, they provide ongoing and valuable insights to the VRUs and partners. A data-sharing platform facilitates the bulk of such data-sharing activities. Many VRUs used visual representations of data through dashboards, including domains such as victims, offenders, types of crime and geographical characteristics. Some reported that this helped demonstrate the value of data sharing to partners, in the sense that there is a lot of data available but the VRU can play a valuable role in collating and codifying this to support each area’s response to SV.

A few VRUs are also aiming to use data to ‘predict’ patterns of SV in the future, thus identifying individuals/groups who are potentially at risk, and informing the targeting of primary and secondary interventions. This requires individual data on a range of risk factors which, as noted, is typically more difficult for VRUs to access and may only be available to a restricted number of VRU staff. One VRU with an advanced dashboard system enabling such monitoring and ‘prediction’, for example, noted that access was limited to just those in the VRU (with police clearance). There is also evidence of few VRUs planning to use advanced statistical methods (including artificial intelligence and machine learning) to focus on preventing rather than reacting to SV trends. However, it was noted that such approaches often require high levels of capacity (including data engineers, analysts, data visualisation specialists), which is not always possible, especially within the short-term nature of the funding period (see section 5.2).

In line with the whole-system approach to respond to issues with effective evidence-based interventions, VRUs reported collecting and monitoring data on interventions they had commissioned, either through internal (management information) systems and/or external evaluations. Such evaluations, however, varied significantly in terms of their focus (covering just process evaluation and/or impact evaluation), as well as their progress in year 2. VRUs collaborated with local evaluators to develop outcome frameworks and key indicators, as well as to track and monitor performance. The following summary box outlines the outcomes and impact indicators VRUs are tracking to monitor and evaluate their performance (some supported more than others through local evaluations):

VRU Combined Outcomes Framework summary findings:

• on average, VRUs are monitoring 30 outcome/impact indicators (499 in total)*
• regarding outcome alignment to the programme-level ToC:
  • 317 (of the 499) were considered ‘outcomes’
  • 60% (of the 317) related to young people and communities, commonly as (school) behaviour and participation in decision-making
  • 40% (of the 317) related to strategic/operational outcomes, including identification of risk factors and hotspots
• impacts aligned to the ToC:
  • 182 (of the 499) were considered ‘impacts’
  • 52% (of the 182) related to reductions in violence, which typically aligned to / expanded on the key Home Office measures
  • 23% (of the 182) related to improved life outcomes for supported young people, including employment
  • 13% (of the 182) related to sustaining a whole-system approach, such as a trauma-informed workforce
  • 12% (of the 182) related to self-reported measures of communities feeling safer

*The analysis excludes only one VRU that has not yet finalised its outcomes framework. It is also important to note that there are commonalities between VRU outcomes where the same or very similar indicators are being used (i.e. there are not 499 unique indicators).

External evaluations have not yet been beneficial for all VRUs. In a minority of cases, local evaluations have not yet produced outputs to inform current interventions, as initially intended. It is unclear when some evaluations will produce findings to inform VRUs at a strategic and intervention level. The causes behind these timelines are also unclear, although some delays were caused in commissioning/setting up evaluations. In some cases, this related to VRUs having bigger priorities (e.g. establishing a comprehensive outcome framework and planning how to use the available data). The process was reportedly delayed even further due to COVID-19. One VRU reported that the method developed for the local evaluation was flawed, which meant further delays in analysing data collected. Since the final product was not completed in time, the VRU is unable to utilise the expected evidence to inform its interventions and projects for the next year. VRUs experiencing these types of challenges are working towards mitigating delays and obstacles, with most of them aiming to publish local evaluation findings within 2021.

5.2 Effectiveness of approaches to data sharing and analysis

5.2.1 Effective data sharing and use

Overall, VRUs have made significant progress in effective data sharing, while still facing some challenges. VRUs have achieved their key data-sharing outputs (as outlined in the ToC) – all have produced SNAs during the first year of funding, while most are using data in their operations and monitoring, and to inform activity. As mentioned, SNAs have largely been considered effective in raising awareness around the use of multiple data sources (as part of a whole-system approach) and establishing a data-driven approach to targeting interventions.

Data sharing case study 1

The VRU’s core team includes a dedicated analyst and an evidence and evaluation lead (two separate posts). The evaluation lead has been central in developing the ToC and an outcome and monitoring framework to inform data collection and analysis. During year 2, the analyst and the evaluator led the development of a Serious Violence Dashboard using live police data to produce weekly, monthly and quarterly reports. On a monthly and quarterly basis, they also receive hospital and other multi-agency data (youth justice, probation, police, education). These regular and detailed outputs enable the VRU to react to emerging issues at a tactical level (e.g. if anecdotally they hear there is an issue, they can use police data to analyse it further) and share data with partners to review and inform delivery.

Quarterly reports are shared at the VRU strategic and operational levels to highlight trends and inform actions. Cohort-level data are also shared with specific partners on an ad hoc basis. For example, it is then possible to do offence-type deep dives that can be shared with the relevant party. Partners receive the findings and analysis, and the VRU organises regular meetings to discuss emerging trends. They also hold individual meetings with the CSPs, with the VRU analyst producing a ‘mini SNA’ for each area to inform discussion.

The VRU is structured to facilitate and strengthen collaboration within the core team, with team members contributing regular input on their areas of expertise. All of this work took place in year 2.

“I don’t just look at the numbers and understand it. Regular input from various different team members about their areas of work have really informed my understanding of the data. For example, when we look at ACEs and risk factors and stuff, we’re not just looking at the numerical data, it’s the qualitative, and having those experts in the team, and being able to connect with partners on a regular basis, and various experts in their areas across the whole county and city, that that has really informed my work.” VRU data analyst

“The data drives our work, so the hotspots that were identified in year 2 are where we are delivering our interventions. So we’re delivering our interventions in the places that need it most, so we’re using data in the right way to make sure that where we’re delivering things is built on a needs based basis.” VRU Stakeholder

VRUs have also shown some early successes in using and sharing data in their operations (e.g. in identifying hotspots), although the vast majority reported that this is still a work in progress and more attention is needed to improve how data are used. Data sharing has also allowed some VRUs to identify gaps in current interventions, in terms of cohorts targeted and people reached.

“There were a couple of things for me that I wasn’t aware of, that were a surprise. It was that 14- to 39-year-olds all have the same likelihood of involvement in violence as a victim or perpetrator. That’s a big broad age range, we’re focusing on 18 to 25 so that was something new.” VRU Stakeholder

Many VRUs reported that sharing and analysing multiple sources of data frequently is key to effective interventions, and, in turn, VRUs’ contribution to preventing and reducing violence. VRU stakeholders also stressed the importance of using hospital data, where available, with police data.

5.2.2 Enablers of effective data sharing

There was broad consensus on factors facilitating effective data and intelligence sharing among police forces:

• multi-agency working, facilitated by the VRU, was reported as key. In particular, the involvement of (public) health colleagues supported (faster) data access

• external support, typically from university departments, has proved effective for several VRUs, given their experience with multiple data sources and how to access these efficiently

• as highlighted in year 1, some VRUs have been able to benefit from wider DSAs already in place (for example, with NHS/hospitals, local authorities or other organisations); existing DSAs and experience of external partners can reduce time and resource requirements (e.g. to set up new agreements) and help facilitate data sharing and analysis; furthermore, the analytical resource provided by the VRU appears to be key in using data effectively; experienced VRU analysts can effectively inform the VRU’s strategic focus, especially when collaborating with a local evaluation lead as seen in a few cases

5.2.3 Barriers to effective data sharing

There was broad consensus across VRU stakeholders on the key challenges faced:

• three main challenges were reported: bureaucratic issues, capacity constraints and difficulty in accessing health data; although VRU leaders acknowledged the importance of data and intelligence sharing, such gaps and challenges are preventing some from effectively using and providing responsive data to inform partners and colleagues

• VRUs have reported that setting up DSAs is a very complicated and time-consuming process; in some cases, disagreement has occurred between partners on what sensitive data (e.g. child exploitation/abuse) they should and should not share; building trust to facilitate data sharing has also been difficult for some VRUs; for example, community organisations have sometimes been less trusting of VRUs, so have not shared data; the short timeframe of the funding period made this even more challenging for some larger VRUs, such as ensuring consistency of data sharing in respect of multiple local authorities

• some VRUs cited capacity constraints due to increasing demand for the VRUs’ work on data (e.g. reports, data dashboards, analyses) in the context of often having only one VRU analyst; the short-term nature of funding has made it difficult to fill posts or retain analysts in some cases, with COVID-19 also leading to redeployments; in many cases, VRUs felt they lacked the resources and internal capacity to ensure the level of analytical capacity and capability required, acknowledging that it would be important to address this

“I think how we do that effectively, and we’re continuing to develop this further, is to make sure that we have invested, and we continue to need to invest in this area, in high quality, not only data, but analysing of that data.” VRU stakeholder

• access to health and NHS data has improved since year 1 but can still pose a challenge for some VRUs; data access was reported as complicated and time-consuming, and hospitals have refused to share data where there are no pre-existing DSAs in place; some VRUs also acknowledged the need to work with partners to improve data consistency and quality; for example, ensuring accurate location data are attached to hospital admissions and consistency across multiple NHS trusts (where they fall into the VRUs geographical remit)

• lastly, COVID-19 has made the analysis and interpretation of data on SV particularly difficult

“COVID-19 is seen as a particular barrier this year on the extent to which conclusions from data this year can be made as it has an impact on almost all outcomes looked at by the [VRU].” VRU Stakeholder

Data sharing case study 2

The VRU commissioned a local university to create a data hub, a tool/platform that pulls together police, fire, health, A&E, and ambulance data (individual record-level data). The university team has been developing such tools for over 10 years so already had many DSAs in place and access to data across the county. They also provide data services to other organisations such as local authorities and are now selling the tool/product to other VRUs across the country.

“I think the reason it works so well this way in here is because the university had already been doing this for a long time prior to the VRUs being set up, so we were already receiving data from A&Es and from the ambulance service anyway. That data was [sic] already being shared in various reports. The data hub allowed us to bring those data sets together, look at them side by side, overlay them and make comparisons.” VRU Operational stakeholder

The VRU analysts and all partners have access to the tool and can run their own graphs and analyses. They predominantly analyse the location and type of crimes from a strategic perspective, not for tactical responses for which the Surge is responsible. The VRU’s aim is to look at the data from the hub to understand the underlying drivers of SV to help embed the whole-system approach, and get partners to look at what drives violence and how it can be prevented, rather than react to specific events.

Work is ongoing to get more data added into the hub, such as from NHS walk-in centres, although these might be at an aggregate level (e.g. LA) and not individual record level. This can help provide some contextual data around risk factors. Access to school exclusion data is seen as another gap at present, alongside NEET data, including from pupil referral units or alternative education providers, to improve cohort targeting.

5.3 Contribution of data sharing and analysis to VRU aims

As outlined in the ToC, data and intelligence sharing is a vital component of the whole-system approach adopted by VRUs. VRUs have shown good evidence of progress against intended outcomes, as SNAs have resulted in data-driven Response Strategies, while partners and stakeholders are generally more aware of, and better understand, the importance of data in SV responses. In some cases, this has directly resulted in improvements in identification and engagement of young people and communities at risk.

While VRUs are not necessarily producing new data (i.e. police and health data existed prior to VRUs), the multi-agency working and capacity provided by VRUs is enabling the sharing and analysis of data together to provide new insights and ongoing, actionable intelligence. VRU stakeholders highlighted that the current level of data sharing and analysis would not be possible without the VRUs. VRU leaders and stakeholders reported the VRU has pushed the creation of new DSAs, promoted collaboration between agencies and partners, and raised awareness about the benefits of using multiple data sources for a whole-system approach. VRUs are thus showing good progress against their specific operational outcomes within the ToC, as partners/professionals understand and are better able to respond to individual/local needs, while the response of the VRU as a whole is evolving and adapting to local needs.

VRUs are also showing an increased interest in data use and sharing, and there was a reportedly high level of awareness and appreciation regarding the importance of data in decision-making and effective response to SV across all VRUs. As many VRU stakeholders have mentioned, one of the most important aims for the future is to use high-quality data and intelligence to be more proactive rather than reactive. As with other core elements of the whole-system approach, data and intelligence sharing continues to be a developing and high-priority area for VRUs.

6. Engagement with communities and young people

This chapter sets out the approaches taken by the VRUs to engage with young people and communities. It also explores the effectiveness of these approaches and how far they have contributed to the ultimate aims of VRUs.

6.1 Approaches to engaging with communities and young people

In year 1, VRUs made good progress in raising awareness of their mission and vision across strategic, operational and delivery partners. However, despite their ambitions, by March 2020, they had made only limited progress in engaging local communities and young people. In year 2, VRUs intended to build on this activity to deliver more diverse and impactful community and young person engagement.

Despite most VRUs having engagement plans in place in year 2, they have made only limited progress in terms of engagement able to inform VRUs’ strategy and objectives. Engagement activities often focused both on communities and young people, without differentiation between the two. Findings are combined for both groups where there is considerable overlap and similarities across them.

6.1.1 Definitions of community

Evidence highlighted that it was not possible to define a ‘community’ using a one-size-fits-all approach. Most VRUs focused their community and young people engagement on specific geographical locations (usually hot spots or areas of risks identified through the SNAs). They emphasised that locality alone did not define their communities and that numerous factors needed to be taken into consideration when understanding local needs and engaging with their target groups (such as different age groups, gender or exposure to risk factors).

“I don’t think that you can probably put one definition on it. I think what you’ve got is you’ve got location, faith, you’ve got different groups of community, I’ve got Muslim groups, Sikh groups, Somali groups that are all their own community within themselves but actually as an area they’re also a community. What we’ve got is different groups working within one locality area, and what we’re trying to do is bring them together.” Strategic lead

6.1.2 Purpose and types of engagement activities

Engagement activities served several purposes. Some VRUs focused on understanding experiences of crime and SV, some aimed to understand communities’ broader needs and concerns, while others planned to co-design interventions with young people and communities. The nature of approaches was also influenced by:

• focus and presence of a dedicated community lead – a dedicated post usually based in the central team responsible for engagement and co-ordination of other partners, VCSE and local activities

• degree of representation of communities and young people on strategic and operational boards to inform the VRUs’ strategy, objectives and engagement plans

• extent of outsourcing – some VRUs outsourced their community engagement to another organisation (usually from the VCSE sector or a CSP) which regularly reported progress to VRUs, who had limited involvement; this approach was common, but not exclusive, to hub-and-spoke models; VRUs adopting it felt local organisations were more embedded in local communities and thus better placed to deliver activities

• existing structures and groups – several VRUs used existing groups (e.g. Youth Parliaments, Youth Assemblies or local youth clubs) to inform and help deliver their engagement strategy; for example, one VRU accessed a national youth network of 84 third-sector organisations involved in addressing youth violence

6.1.3 Examples of community and young people engagement

VRUs adopted many approaches to engagement, ranging from sending out newsletters to co-production of interventions and participation in commissioning of services. The most frequently mentioned approaches can be categorised as follows:

• provision of communications, newsletters and campaigns – including social media, press or radio campaigns aimed at different target audiences; campaigns usually focused on a specific issue rather than promoting or raising awareness of the VRU, with some co-designed by young people

  “We got young people from alternative provisions, so very much at-risk youth who were very much engaged with agencies or community organisations, and sat them in a room and said, ‘What would literally stop you in your tracks? What would change your behaviour? What would make you stop and think?’ And from that we got some quite raw audio and footage of those young people talking about their experiences with serious violence and growing up specifically in [VRU Area] and we’ve been able to pull out this amazing social media campaign over a number of weeks and the reach of that has been absolutely phenomenal.” VRU lead

• holding community events and consultations, some aimed at specific audiences, while others focused on wider engagement in a specific area; in year 2, most engagement was facilitated online due to COVID-19

• use of digital platforms and tools to ask questions and gather feedback on concerns or local issues and invite views on possible solutions and their potential effectiveness

• facilitation of educational talks on topics particularly prominent locally –usually delivered as one-off talks by the voluntary sector, youth workers and/or the police, or through assemblies in school

• use of Young Ambassadors, Youth Parliaments or Community Action Groups as a vehicle and focal point for young people and communities to have a voice, act as a sounding board, and often contribute to co-design

• use of surveys or focus groups to explore perceptions of community safety, identify local concerns and needs, and understand experiences in relation to SV. Some VRUs used the results to inform the review of their initial SNAs, while others were still deciding how to best use the insight provided

• gathering feedback from young people, especially the identified at-risk cohorts, through commissioned interventions – feedback was either shared with the VRU by the organisations delivering interventions or collected directly by the VRU; feedback was sometimes used to shape the content or delivery of a specific intervention or to highlight specific concerns related to the local area

• outsourcing commissioning of interventions to local community-based organisations with the proviso that young people and communities were involved in co-designing the activities

Annex 5 includes several detailed examples of the range of engagement activities delivered by VRUs.

6.1.4 Effectiveness of approaches to engagement with communities and young people

While VRUs’ aspirations were high at the outset of year 2, progress in engagement was somewhat slower than expected. Despite limited progress, nearly all VRUs had started delivering their intended engagement plans. Initial evidence from a few VRUs suggested that young people and communities were engaged or had started engaging with VRUs and partners, and that some were changing their attitudes towards the police or statutory services as a result. Evidence from a few interviews with young people involved in the co-design of interventions, or who were directly informing VRU strategic plans, suggested that they were feeling empowered and listened to.

“They’re giving us multiple opportunities and they’re providing us opportunities that we can pass onto our own people, our own community, our friends. The group allows us to have our own voices, and it allows our voice to be heard, and we’re just representing our public. It gives us, they let us have our voice, and they take what we say and they actually do something about it. Like we’re making a big impact.” Young person, Advisory Group member

6.1.5 Enablers of effective young people and community engagement

Despite the limited progress and evidence available, there were some emerging trends that highlighted key enablers for effective engagement:

• existing relationships and established routes – VRUs with more established routes to access communities and young people had made more progress; many VRUs discussed the need to take the time to build trust and a strong rapport with communities to harness the buy-in required

• dedicated community roles – several VRUs noted that having a dedicated community role based in the central team had been integral to pursuing their engagement plans

• using young people and communities to co-create content and campaigns – some VRUs felt this made content more relatable and accessible; however, limited evidence was collected to enable an assessment of the impact of the campaigns on local communities

A few VRUs worked with local communities in the wake of shock and global events (such as local murders or the Black Lives Matter movement) to harness their views. Although not seen as an enabler of engagement, it helped to activate the community and break down barriers between them and the police.

6.1.6 Barriers to effective young people and community engagement

VRUs faced several barriers to effective engagement, which were often outside of their control. The main barriers to engagement were:

• COVID-19 – the global pandemic created a myriad of issues:
  • the pandemic put additional pressures on VRUs, strategic partners and frontline organisations; many had to prioritise responding to the crisis due to lack of time and resources, affecting engagement activities
  • the pandemic put pressure on communities and young people themselves, creating further barriers
  • nearly all face-to-face engagement was cancelled, requiring VRUs to develop alternative methods that took time to develop and had varying levels of success
• lack of resource – a minority of VRUs reported that they simply had not allocated sufficient resource to lead their engagement activities; they therefore planned to create dedicated posts in year 3

“We haven’t had a consistent approach across [area] and that is mainly down to lack of resource in the core team. So that is an issue that I’ve tried to address with the funding for next year, which is for us to have a community consultation and engagement post. Someone who can … develop some sort of strategy for community engagement and consultation right across [area]. That is something that has been missing.” VRU strategic member

• diversity – several stakeholders noted communities were very diverse, so designing the right approach was therefore challenging; what worked for one group within a community might discourage others; similarly, a few stakeholders noted it had been very challenging to recruit a diversity of representatives and ensure that the hardest to reach groups felt listened to

• negative perceptions towards police and statutory services – despite some early changes in perceptions of the police and statutory services, negative attitudes remained evident; a few stakeholders acknowledged it was important to have police present and engaged in community activities, but perceived to be led by community or local organisations to help address the perception related barriers

• limited engagement across the different levels of the VRU – some interviewees noted that while their lead and central team were aware and engaged with the relevant community and young people workstreams, awareness and engagement across the different strategic and operational levels varied; for example, a few frontline staff and external providers noted that they only had a very high-level understanding of what the VRU was trying to achieve in terms of engagement

• top-down approach – a few frontline workers noted some VRUs focused on designing a strategy first rather than making sure that communities were involved in co-design at the outset; this could result in the strategies not meeting the objectives of engagement

“At the moment it feels like it’s first strategy and then delivery plan, rather than the other way round and I think they’ve missed a trick there. That community led experience is missing.” Frontline worker

6.1.7 Additional learning for future engagement activities

Early evidence also showed that nearly all VRUs agreed that while some online approaches should be adopted and considered in the future, they cannot fully replace face-to-face contact, particularly when engaging the most at-risk communities and young people, where challenging negative perceptions and tackling mistrust of statutory organisations was key.

“We wanted to get a real cross-section of young people involved, so I think initially our ambition was to have some young people who were in the youth justice system, people who had perpetrated violent offences. We didn’t just want it to be your high-achieving bright kids from middle class backgrounds, we wanted a real good cross-section. I think to a large degree we’ve achieved that. I think it’s probably been more difficult to do than if we’d been able to do face to face work.” VRU member

In addition, several stakeholders also questioned the extent to which it was necessary to associate engagement activities with the VRU brand, which could be negatively perceived and deter people to get involved. As a result, some VRUs were purposefully not using their branding on campaigns and materials, and chose to be represented by local or delivery organisations instead.

6.2 Contributions of engagement with communities and young people to VRU aims

Nearly all VRUs acknowledged that progress towards community and young people engagement was limited in year 2 and remained a key priority for year 3.

“I think if [name] and I put our hands on our hearts, I think we need to do more around community engagement and hearing the voice of communities. I think it’s been challenging for us and potentially one that has not sat at the top of our priority list. I think that really needs to be addressed next year.” VRU lead

As a result, limited evidence was available to enable an assessment of the extent to which VRU engagement activities had contributed to the achievement of the programme outcomes. For example, while there were some examples of VRUs’ positive engagement, limited evidence outlining any progress made towards the key performance indicators aiming to achieve increased perceptions of community safety was available. Where evidence was collected, this often related to perceptions of the interventions or to help inform understanding of local needs. In addition, in some cases where successful engagement was reported, stakeholders noted they had not had a chance to fully analyse or understand the feedback collected and decide on how to use it. Stakeholders also noted that it was too soon to judge whether their overall engagement strategy was effective, as they could only talk about short term or immediate outcomes and their sustainability was not clear.

Greater progress is therefore anticipated on this theme during year 3 of the programme, which is hoped will be less disrupted by the ongoing pandemic.

7. Interventions

This chapter describes the approaches taken to commissioning interventions. It also explores their effectiveness and outlines the extent to which they have contributed to the achievement of the ultimate aims of the VRUs.

7.1 Approaches to the commissioning of interventions

In year 1, VRUs commissioned their suite of interventions prior to the development of the SNAs and associated Response Strategies. As such, they had limited opportunities to embed a strategic approach into commissioning processes. Commissioning, therefore, varied widely across VRUs and tended to be more reactive in nature.

Evidence from year 2 suggested commissioning had evolved to become more strategic, and to some extent more evidence-based, across VRUs. This included most VRUs becoming more purposeful in their commissioning approach, with 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.

7.1.1 Strategic level

VRUs used a range of evidence to better understand local needs and inform the portfolio of interventions commissioned. The majority used their SNAs and Response Strategies as central pieces of evidence in year 2, and the remaining few had intentions to do so following more comprehensive development of their evidence base. These provided a deeper understanding of at-risk cohorts, existing services and gaps in provision, which in turn formed the basis for commissioning.

“Each of the interventions that we have commissioned can be mapped back to the Strategic Needs Assessment, and where something has been cited as good practice, or there’s evidence, it could be emerging evidence, but evidence of it working.” Strategic Stakeholder

Several VRUs also used wider external evidence sources to inform commissioning. This included the evidence base on the Scottish VRU, influencing the adoption of some of its flagship interventions, e.g. Mentors in Violence Prevention programme and A&E Navigators across several VRUs. It also included evidence drawn from Early Intervention Youth Fund or Youth Endowment Fund (YEF) reports and, in a minority of cases, the academic evidence base built in the USA.

A minority of VRUs also drew on the following types of evidence to inform commissioning:

• learning and early findings from local evaluations shared across VRUs; one VRU, for instance, shared evidence of its domestic violence perpetrator research through a regional Violence Reduction Network

• in-house or commissioned research, especially to fill gaps or add more depth to the evidence provided in SNAs; one VRU reported this had enabled them to continuously develop their understanding of target groups

“The SNA identified a need to refresh our understanding of the experiences of serious violence for girls and young women. Rather than commission a research partner to carry out qualitative research, the VRU has funded local partners, including those in the community sector, who are already working with girls and young women to gather their views, and better understand their personal experience of violence.” Strategic Stakeholder

• outputs from consultations and/or co-design with community partners were also used to influence the type and location of commissioned activities

7.1.2 Operational level

A variety of commissioning approaches were observed, which appeared to be partly driven by the model of working adopted by the VRU. This included the adoption of a more decentralised approach to the commissioning of interventions by several VRUs, particularly where they covered heterogenous areas which served diverse communities. A decentralised approach was also perceived to increase local stakeholder buy-in to VRU objectives, by offering partners the opportunity to influence the commissioning process. This was especially evident where VRUs adopted a hub-and-spoke model, or in those centralised VRUs that operationalised their activities through a series of workstreams or through ‘quasi-hub-and-spoke’ models (and in some cases, both).

Several VRUs also adopted a scaled approach to commissioning, where area-wide / large-scale interventions were commissioned centrally, and smaller, more community-based interventions were commissioned locally. This was thought to help retain some strategic consistency within whole-VRU flagship interventions, while ensuring local needs were being met via a more local commissioning channel. The establishment of shared outcomes frameworks, used as the basis for all commissioning, were employed to ensure alignment between these different approaches in some cases.

“We have two methods to our delivery. We have VRU from the direct programmes, which have to be pan-[VRU], which means if someone wants to bid or if we’ve got a programme, it’s got to be a pan-[VRU] bid for us to take it here. We can’t have a bid coming in from one area for one local authority. Anything for one authority, we pass that to the city partnerships to deal with.” Operational Stakeholder

Linked to the point above, governance processes for commissioning tended to be proportionate to the scale of the funding. Larger-scale interventions – usually above a fixed funding threshold – demanded higher levels of scrutiny, while smaller ones were commissioned with more limited engagement from the governance boards. Table 7.1 presents the funding spent on interventions across VRUs in year 2. This shows considerable variation as each VRU was given the autonomy to determine how to allocate and prioritise how they spent their allocation. There is a general pattern of VRUs that received higher funding allocations spending a greater proportion on interventions. Further research is required to understand this pattern, which is likely to result from a combination of factors potentially including the fixed operational cost associated with delivery of a VRU being only marginally different between varying sizes of VRUs, potential match funding, and the impact of COVID-19 in year 2.

Table 7.1: VRU intervention funding allocation

VRU	Funding allocation	Intervention spend total	% of total funding
London	£7,000,000.00	£6,768,000.00	97%
Northumbria	£1,600,000.00	£1,108,210	69%
Merseyside	£3,370,000.00	£2,328,237	69%
Greater Manchester	£3,370,000.00	£2,171,521	64%
Sussex	£880,000.00	£549,476	62%
West Midlands	£3,370,000.00	£2,069,154	61%
Thames Valley	£1,160,000.00	£692,709	60%
West Yorkshire	£3,370,000.00	£1,985,579	59%
Hampshire	£880,000.00	£487,484	55%
Kent	£1,160,000.00	£633,043	55%
Avon and Somerset	£1,160,000.00	£604,290	52%
South Yorkshire	£1,600,000.00	£759,316	47%
South Wales	£880,000.00	£414,650	47%
Essex	£1,160,000.00	£524,700	45%
Bedfordshire	£880,000.00	£390,115	44%
Lancashire	£1,160,000.00	£469,405	40%
Leicestershire	£880,000.00	£341,358	39%
Nottinghamshire	£880,000.00	£332,673	38%

7.2 Types of interventions commissioned

In year 1, time constraints and the need to commission interventions prior to the development of SNAs and Response Strategies led to funding being largely directed towards extending and enhancing existing interventions. Consequently, VRUs recognised that their intervention portfolio was likely to be subject to significant change in year 2, to ensure it reflected the needs identified in their SNAs and associated Response Strategies.

A review of year 2 funding applications (specifically, delivery plans, which varied in comprehensiveness and quality), monitoring information and evidence from the qualitative interviews indicated the following delivery intentions:

• the 18 VRUs planned to commission a portfolio of at least 216 interventions (see Table 7.2); just over half of these interventions intended to target young people that were already involved in crime/violence or those who had a known risk or were known to services (27% and 28% respectively), and one in seven (14%) intended to target those who were potentially high risk; the majority of the remaining interventions (30%) were to be universal in their nature

• a programme-level comparison between interventions planned in years 1 and 2 indicated an intention to place more of an emphasis on young people in levels 1 (universal) and 4 (involved in crime/violence) in year 2; however, this finding should be treated with caution; different data collection methods were used in years 1 and 2, and this form of comparison does not recognise the difference in scale and reach of each of the intended interventions

Table 7.2: Number of planned interventions by VRUs in year 2 by target groups

Target groups	Year 1 planned interventions Number	Year 1 planned interventions %	Year 2 planned interventions Number	Year 2 planned interventions %
Level 4: Involved in crime/violence	35	20%	58	27%
Level 3: Known risk/to services	48	27%	60	28%
Level 2: Potentially high risk	51	29%	31	14%
Level 1: Universal	34	19%	64	30%
Other	7	4%	3	1%
Total	175	100%	216	100%

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

• in relation to delivery models, 32% (69) were intended to be preventative interventions, 23% early interventions, 20% (43) therapeutic interventions, 24% (52) desistance interventions, and 1% (2) were unknown or unclassified; table 7.3 below shows the relationship between the delivery models and the target groups; there is evidence of a (expected) relationship between delivery models and the target groups, with therapeutic and desistance interventions more common for target groups known to services or involved in crime/violence, and prevention /early intervention more common for the universal target group

Table 7.3: Number of planned interventions by VRUs in year 2 by target groups and delivery models

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

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

• an intention to focus most interventions on those under 25, with many covering multiple age groups, and a smaller proportion of funding directed toward over 25s (e.g. parents, ex-offenders, wider community members), whole families or communities

• commissioned interventions were likely to be split equally between pan-VRU and area- or setting-specific geographies, with 47% (102) being pan-VRU and 53% (114) based in specific areas or settings (hospitals, schools etc)

• thirty-five percent (75) of interventions were to be commissioned to VCSE organisations, 25% (53) to local authorities or services (including youth offending and victim support), 6% (12) to the police/HMPPS, 5% (11) to NHS Trust/health system and 3% (6) directly to schools; it was unclear what type of organisation was intended to be commissioned to deliver the remaining 27% (59) of planned interventions

• forty-two percent (90) were intended to be interventions not delivered in the area previously, while 37% (79) were expansions of existing delivery to reach new locations, new cohorts or both; around 19% (42) of interventions were continuations from year 1 and 2% (5) were unknown or unclassified

• only 5% (10) were intended to be new intervention models not delivered before (in any area); twenty-seven percent (58) were reported to be evidence-based, while for the vast majority (69%; 148) this information was unknown; for those interventions that were evidence based, 43% (25) were based on the VRUs’ own evidence, and 57% (33) were based on wider external evidence

Although it has not been possible to provide a comprehensive comparison between year 2 intended and actual delivery due to the incomplete nature of the planned intervention data, validated monitoring returns from 17 of the 18 VRUs showed that 331 interventions were delivered (see Table 7.4). Over half of interventions targeted young people already involved in crime/violence or those who had a known risk or were known to services (20% and 35% respectively), and one in seven (14%) targeted those who were potentially high risk. Most of the remaining interventions (26%) were universal in their nature. Comprehensive comparable delivery data for year 1 was not available.

Table 7.4: Number of interventions delivered by VRUs in year 2

Target groups	Number of interventions delivered
	Total	%
Level 4: Involved in crime/violence	67	20%
Level 3: Known risk/to services	116	35%
Level 2: Potentially high risk	45	14%
Level 1: Universal	87	26%
Other	16	5%
Total	331	100%

NOTE: Data are based on validated monitoring returns from 17 of 18 of VRUs.

7.3 Effectiveness of approaches to commissioning interventions

Overall, there was some evidence to suggest that VRUs had made positive progress towards delivering more effective commissioning. Approaches became more strategic as VRUs incorporated learning from year 1, using SNAs and Response Strategies to inform decision-making. Flexibility offered by the Home Office to embed localised and context specific commissioning was also key to ensuring VRUs could respond to COVID-19. In a few VRUs, commissioning frameworks and/or alignment with Surge also supported effective commissioning.

7.3.1 Enablers of the effective commissioning of interventions

Several key enablers led to more effective commissioning:

• support from core/central VRU teams was vital to operationalise VRUs’ strategic vision, in turn enabling more targeted commissioning; this was especially important for VRUs that decentralised commissioning; the following VRU-wide, cross-cutting pieces of work proved particularly useful:

  • mapping exercises were being, or had been, undertaken by most VRUs to review existing interventions at strategic, operational and local levels to enhance efficiency, reduce duplication and fill gaps

  • in combination with mapping exercises, SNA and Response Strategies helped to identify gaps in provision and focus conversations with local operational staff; in some VRUs, strategic or senior operational staff (e.g. workstream leads) held workshops with operational staff involved in local commissioning (e.g. spokes, CSPs) to present evidence on the local area (from data tools/analysis, SNAs and mapping exercises) and discuss priorities for commissioning in light of identified gaps in provision; some VRUs also discussed issues and priorities with their local operational staff through more ad hoc and informal conversations

• some VRUs formalised, or were formalising, commissioning by developing commissioning or contractual frameworks; they felt this enabled a fair, consistent and transparent approach, perceived as being increasingly important when decentralising commissioning; in several cases, frameworks aligned with SNAs, Theories of Change, and/or evaluation or outcome frameworks; this was felt to ensure all interventions worked towards the VRU aims and objectives, and to enable consistent monitoring and evaluation

• in some cases, VRUs focused on ensuring that Surge and VRU funding complemented one another; strategic and operational VRU staff perceived this alignment positively; for example, one VRU focused this alignment on hotspot areas, targeting preventative work in locations where Surge operations also took place

• VRUs appreciated the flexibility offered by the Home Office in relation to commissioning, especially as COVID-19 required them to amend their plans in response; some VRUs re-profiled commissioning to tackle needs emerging from the pandemic (e.g. the surge in domestic violence cases), acknowledging that this deviated from initial SNA and Response Strategies (and sometimes SV definition); in others, commissioning plans remained unchanged, but some interventions were postponed until later in the year when delivery was felt to be more appropriate

• in two VRUs, match funding was committed to support commissioning over a two or three-year period, helping to embed longer-term commissioning

7.3.2 Barriers to the effective commissioning of interventions

Several challenges and barriers were identified:

• some VRUs had not yet struck the appropriate balance in relation to oversight of their commissioning, which was felt to be problematic and requiring further consideration; several opted for a ‘proportionate governance’ approach, with governance boards retaining at least some oversight over the commissioning decisions of the workstream leads or local spokes; however, this was not always the case; at least one VRU acknowledged having no oversight of CSPs’ local commissioning; conversely, in another VRU, scrutiny from governance boards (the Office of the Police and Crime Commissioner (OPCC) in particular), and long commissioning processes for large-scale interventions, meant the VRU deliberately focused on commissioning small-scale activities to avoid this

• nearly all VRUs reported that the timing and timescales for commissioning in year 2 had been a challenge; the development of the SNAs and the Response Strategies ran in parallel with the commissioning timeframe at the start of year 2, creating some capacity constraints for core VRU teams; this led some VRUs to be less strategic in commissioning than intended, given SNAs and Response Strategies were still being developed

“With the other [interventions], it was quite a quick and dirty exercise, with the Police and Crime Commissioner’s office, looking at what exists, what’s delivering well and what could we be funding using that money. I think, in an ideal world, we would have taken a bit longer and have had more of an opportunity to reflect on that, but the timescales were not ours. The Home Office gave us a really tight timescale for turning this around.” Strategic Stakeholder

• VRUs also stated that the short-term nature of funding prevented them from adopting the long-term approach to commissioning required to deliver their aims and objectives

• the COVID-19 pandemic led several VRUs to delay commissioning, causing an underspend towards the end of year 2; however, as mentioned, this sometimes enabled VRUs to meet shorter-term needs and increased their flexibility to respond to challenges raised by COVID-19

• in some VRUs, efforts were still needed to operationalise their strategic vision efficiently, such as ensuring that knowledge of existing/commissioned services was better shared with operational staff to avoid duplication when commissioning at the local level; this was particularly the case where VRUs had decentralised commissioning or adopted a hub-and-spoke model

Evidence also highlighted some learning and considerations for future commissioning. Some VRUs noted the need to better understand crime flows and networks to inform commissioning, rather than simply looking at ‘hotspot’ areas. For some crimes, particularly those related to county lines or the night-time economy, victims and perpetrators sometimes came from areas outside the VRU remit to commit crime in particular hotspots. In this scenario, commissioning interventions in the hotspot area alone may not be sufficient. Ways to address this could include increased sharing of granular, especially individual-level, data within and between VRUs to enhance understanding of at-risk cohorts/individuals as well as crime flows and networks. The next iteration of the SNAs and Response Strategies could incorporate this, once data-sharing processes are more developed.

7.4 Effectiveness of delivery of interventions

With an estimated 262,210 ^{[footnote 17]} individuals supported in year 2, there is early indicative evidence to suggest that VRUs had effectively reached children, young people and adults through the delivery of interventions. Evidence suggested that more progress had been made in VRUs that had existing interventions or partnerships with organisations with a ‘proven’ delivery model in place – including more established services and referral pathways. We cannot comment on the effectiveness of the individual interventions delivered at this stage, as most 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.

7.4.1 Enablers of effective intervention delivery

At the time of reporting, an analysis of the monitoring information available indicated that interventions had reached 262,210 individuals across 17 of the 18 VRUs (see Table 7.5). While direct comparison between year 1 and year 2 data is not possible due to differences in the data collection methods and year 1 data being incomplete, it suggests that more young people were likely to have been reached in year 2. This appeared to have been driven by a significant increase in the numbers in the level 1 universal target group being supported in year 2 relative to year 1.

The evaluation 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). In the small number of cases where intended targets had been set, these had become redundant because of the COVID-19 pandemic.

Table 7.5: Number and proportion of individuals supported by VRUs in years 1 and 2 across target groups

Target group	Known individuals reached in year 1* - No.	Known individuals reached in year 1* - %	Total individual reached in year 2 - No.	Total individual reached in year 2 - %
Level 4: Involved in crime/violence	7,028	7%	6,412	2%
Level 3: Known risk/to services	9,746	9.5%	18,847	7%
Level 2: Potentially high risk	34,323	33%	18,698	7%
Level 1: Universal	50,826	49%	214,561	82%
Other	1,455	1.5%	3,692	1%
Total	103,378	100%	262,210	100%

NOTE *Year 1 data are based on information provided by 175 of the VRU interventions, which did not represent the totality of all that were commissioned.

Despite limited progress and available evidence, there were some emerging trends that highlighted key enablers for effective intervention delivery, including:

• in a few cases, in-house delivery teams within VRUs appeared to have been an effective way to sustain delivery throughout the uncertainties of COVID-19

• considering the profile of the organisations commissioned, with local organisations seen as better equipped to understand and meet the needs of their communities, for example by being able to engage target groups through established referral pathways; however, it was noted this type or organisation often lacked capacity to implement monitoring and evaluation activities and needed more VRU support in this regard

• interventions led by delivery staff with lived experiences of SV (either as a victim or perpetrator) worked well to engage participants, as young people could better identify and relate to them; however, this finding should be treated with caution; although it was noted in wider literature that hearing testimonials of individuals with direct (‘lived’) experience of knife crime can be impactful (Foster, 2013), other evidence relating to using testimonials from ex-drug users has also been associated with no or negative prevention outcomes (Warren, 2016)

  “We found in those interventions that did not have some people with lived experience, the engagement rates aren’t as high. Those who have previous experience like YOS bring with them an energy and ability to engage with young people, the amount of the young people I spoke to who were saying, ‘They’re just amazing, they’re much better than my [statutory service], never had a relationship like this, they genuinely care.’” Operational Lead (Evaluation)

• in limited cases, the effects of COVID-19 positively influenced delivery; this was usually the case for training programmes with professional staff, where online delivery meant that more people could participate

• to support the delivery of interventions, stakeholders perceived the VRUs had played a central role in:

  • embedding monitoring and evaluation plans throughout commissioned interventions and, in some cases, building the capacity of frontline and external providers; this was achieved in different ways across VRUs, including the development of commissioning, outcomes and evaluation frameworks, ToCs and toolkits shared with providers/bidders; most VRUs had also begun to build their own evaluation capacities (to varying extents), procuring those services either externally or embedding them internally
  • encouraging knowledge sharing within VRUs, including between intervention providers; this support varied in nature; some VRUs set up forums and/or knowledge exchange channels for providers to connect and network, share delivery updates, gain an oversight of other services delivered, and exchange knowledge and ideas; in other VRUs, stakeholders felt such forums and knowledge exchange channels were too limited, and they encouraged the VRUs to take an enhanced role in this arena; one VRU set up named contacts in the VRU team responsible for their different localities, an approach appreciated by frontline staff as they could build a rapport with one person and felt they could raise emerging findings and challenges and get advice; they also set up a Knowledge Hub to centralise and disseminate information, particularly regarding standards for delivery (e.g. safeguarding) and monitoring and evaluation practices
  • in a few VRUs, there was evidence that discussions at forums, or evidence from monitoring and evaluating interventions, had informed subsequent delivery or commissioning; however, examples of such feedback loops were limited; for the majority of VRUs, monitoring and evaluation was still ongoing, and VRUs were not in a position to implement feedback loops

Box 7.1: Example of successful embedding of monitoring and evaluation plans across interventions

VRU model: Centralised

Context: VRU A was originally set up as three-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 2 for the next three years to ensure that learnings are shared and taken on board throughout the year.

Evidence indicated that the presence of local, in-house evaluation capacity was key to developing a holistic approach to evidence use in this VRU, helping ensure effective commissioning and delivery. Local evaluation also helped embed monitoring and evaluation practices across all interventions. The local evaluator set up ToCs, outcomes frameworks and monitoring forms for each intervention, aligning these with the VRU objectives to capture the contribution of each intervention. A commissioning framework required providers to commit to ongoing feedback. The local evaluator also had a central role in building the capacity and confidence of local partners to engage in their own evaluation activities, in response to delivery organisations requesting advice on designing their own ToCs and measuring outcomes.

The VRU also established a strong feedback loop through the outcomes framework and monitoring forms. Data from interventions was regularly analysed and fed back to the VRU lead and to the board and partners/frontline staff, where applicable. This enabled the VRU to identify early outcomes and emerging issues, so as to ‘act quickly’ to address any challenges that emerged. Where certain topics or approaches proved effective, intervention staff were encouraged to focus on that aspect of their delivery.

The recruitment of a VRU evaluation lead over a three-year period also helped the VRU to take a longer-term approach to developing the local evidence base, ensuring that it would inform future commissioning decisions. In year 2, the VRU commissioned a mix of new and existing interventions, with this also being grounded in wider evidence from other similar programmes and the literature. Going forward, the local evaluator noted they will always start with a review of the evidence available before considering commissioning a service, which will include increasingly drawing on their own evidence base.

7.4.2 Barriers to effective intervention delivery

VRUs had all made some progress to set up and deliver their planned interventions; however, delivery progress was more limited than anticipated for several reasons:

• the outbreak of the COVID-19 pandemic in March 2020 negatively affected the capacity of providers to deliver interventions in most VRU areas; some had to pivot activities to provide support to communities to meet more immediate needs (e.g. door-to-door checks, safeguarding referrals, foodbanks or food deliveries); planned capacity from existing services (e.g. police, health professionals) was often redeployed – for instance, an intervention aimed at preventing domestic abuse had to be postponed as the Independent Domestic Violence Advocates (IDVA) service normally present in police control rooms was working remotely as its staff were civilians

• furthermore, the combination of national and local lockdowns and associated social distancing restrictions meant that most planned delivery had been delayed, rescoped or adapted; where appropriate, VRUs had found innovative alternatives and embraced virtual and digital modes of delivery. VRUs and frontline/external providers reported having learnt a great deal throughout the COVID-19 pandemic, which resulted in them being more prepared for the second and third national lockdowns; however, it had not been possible to migrate many planned face-to-face interventions to virtual delivery modes for a variety of reasons:
  • accessing technology can be an issue – for example, a single-parent family with more than one child involved in an intervention with the only device that could be used being the parent’s smart phone
  • recruiting, engaging and retaining participants online is much harder, with dropout rates found to be higher
  • building trust and rapport with harder to reach groups or most at-risk cohorts is often much harder online
  • ensuring those supported could access a safe space to engage in virtual activities also posed challenges

• VRUs consistently reported the short-term nature of the funding cycle as a barrier to sustaining engagement of providers and participants; providers were less likely to continue to recruit participants towards the end of the intervention period, as they were uncertain they could offer the consistency of support needed; providers were also concerned about securing other funding to retain staff should VRU funding be discontinued

• the slower pace (than anticipated) of delivery had a knock-on effect on monitoring and evaluation; most VRUs planned monitoring and evaluation of their interventions, but these were delayed in line with delivery; stakeholders also mentioned that COVID-19 had made monitoring and evaluation challenging; this meant that local evaluations were ongoing at the point of drafting this report, and monitoring data was limited across VRUs

Evidence highlighted some learning and considerations for future intervention delivery. The evaluation noted significant disparities in monitoring and evaluation capacities and practices across VRUs. A minority had made significant progress in embedding monitoring across interventions, for example agreeing KPIs. Transfer of this and other good practice to enhance intervention monitoring and evaluation across all VRUs would be beneficial.

7.5 Contribution of interventions to VRU aims

All VRUs made some progress in setting up and delivering interventions. However, it should be noted that COVID-19 and the effects outlined meant that, overall, there was limited evidence to comprehensively assess their contribution to the achievement of the VRUs’ aims and objectives. Despite this constraint, there is emerging evidence that some VRUs made positive progress towards the achievement of a range of shorter-term outcomes through their interventions that began contributing to some longer-term programme-level outcomes. Short-term outcomes achieved include: improved identification and engagement of target groups; improved referral of young people to interventions; increased engagement among young people; and improved attitudes towards statutory services or willingness to communicate and engage with services or providers.

In addition, the small number of young people interviewed all spoke positively about their engagement in interventions ^{[footnote 18]}. However, none were aware or had heard of the VRUs, except one individual who had participated in the relevant VRU’s youth engagement activities. Given the nature of the interviews conducted with young people ^{[footnote 19]}, it is not possible to draw firm conclusions on outcomes achieved by different types of interventions. However, the evidence collected did provide early indications of what had worked well for the group of young people interviewed. This included an appreciation of the informality of sessions/activities; a preference for one-to-one and, where possible, face-to-face activities; and the importance of tailored and relatable content.

“To be honest, I think people will benefit from it if they can relate to it. This group that was there, they could relate to what was going on in the course. Just all jumping at it with our opinions on certain things. We could all relate to it, you know what I mean? I think if people can relate to the course, they benefit it from.” Young person

There were further early indications that the VRU interventions were beginning to contribute to some of the longer-term outcomes set out in the programme’s ToC around the improvement of life outcomes for young people. This included outcomes that related to:

• improved confidence and self-esteem
• improved motivation/positivity/resilience and overall wellbeing, including mental health
• improved relationships with family, friends and others

“I see a bit more of my dad. I speak to my sister a bit more. He’s like, ‘Is it worth keeping that grudge or is it better just to forgive and forget? He was like, ‘If you can’t forgive me, you can’t forget. You can’t forget when you can’t forgive.’ I have maybe not fully forgiven him but I’m not angry with him anymore, do you know what I mean? I can talk to him now but, last year, I’d have punched his head in if I’d have seen him.” Young person

• movement towards the reduction of young people NEET

“I am grateful for the [X] group, honestly working with [Y] has helped me find opportunities, he would talk to me about different help I could find and other paths I could explore, he has helped me look for different educational paths and then helped me find a job … I will keep up this positive change and am so happy I am now making an honest living. Thank you [Y] and the group.” Young person

• improved behaviours (e.g. anger management)

“It made me think a lot on my actions. It makes me think before I actually act on anything, before I speak … I kick off over the slightest thing basically. Yes, the course, it has made me think a couple of steps ahead, like is it even worth it? If you do that, then you’re going to do this … It has made me more aware with myself, my surroundings, how I am. How I act and respond to other people.” Young person

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

“I reckon if I had just stayed stubborn like I used to be and I didn’t come back to the second [session], to be honest, I think I might have ended up back in jail … I would have definitely got back into trouble. Not meaning to but just not thinking straight. I would just go straight in there. I’d just go and do it.” Young Person

The vignette in Box 7.2 below provides a case study based on our interview with one young person.

Box 7.2 Case study: Transforming Life Choices (based on an interview with one young person)

Tom (name changed) took part in the Transforming Life Choices programme aimed at improving wellbeing and mental health to enable better life choices. Over 12 to 15 weeks with four other young people, he engaged in weekly sessions involving scenario-based activities and discussions. In the last session, participants were asked to prepare a piece to present back to the people that had referred them (e.g. friends, family, support organisations), reflecting on what they had learnt though their participation.

Tom was referred by his support worker as he was going through a difficult period. While he initially had no expectation about the programme, he felt in hindsight that he had positively benefitted and would recommend it to others.

“I honestly wasn’t too sure what it was about … At that point, there was a lot of support being offered to me and was just taking everything but I wasn’t too sure what to expect and I thought if there’s a possibility it’s going to help me, I’ll take it. Because, I wasn’t feeling great at the time so I thought if it will, I’ll take it. I’m so glad that I did and I’m actually looking to volunteer with them in the future and their future programmes and things like that because it was so helpful for me.” Young person

Tom indicated that the programme had helped to improve his confidence, self-esteem and mental health by encouraging him to take a more positive viewpoint on life situations and events. He also mentioned that the programme had helped him to better manage his feelings and behaviours.

“[T]he people that know me have seen that I changed as a person … it’s made a massive improvement on me as a person and I do now look at, well, I do obviously still see negatives in things quite a lot. But I do see a lot more positives whereas before I didn’t see any positives or anything at all.” Young person

“It teaches you not only to feel better about yourself but also to control yourself and your thoughts and feelings when things don’t really go the way that you wanted them to go. I say like losing a game or if you’ve had an argument with someone, rather than just screaming and shouting at them to then speak and say, ‘Well, we can sort it out and talk about things.’ Rather than just being, ‘Oh, I hate you, you’ve done this, that,’ and storming off.” Young person

Finally, several stakeholders also noted that disaggregating the contribution between the VRU and Surge funding and activities was particularly difficult where the two streams of funding had been strongly aligned. They also mentioned that it was likely that the Surge activities would have more impact in the short term.

“At the moment, it’s caveat. It’s just saying to people look, we’re seeing good stuff. We don’t know if it’s us. (..) I’m always inclined to actually point at Surge and go. I think these guys are the heroes not us. We come later, we will be heroes later.” Operational Stakeholder

In summary, VRUs developed SNAs, which enabled the identification of primary at-risk cohorts and the hotspots (and hot times) where incidents were most likely to take place. The SNAs provided an initial evidence-based foundation upon which resultant Response Strategies were developed. Although variable in depth and quality, these informed the focus and activities of the programme, including the portfolio of commissioned interventions.

VRUs adopted a variety of commissioning approaches, including both centralised and localised. This led to the commissioning and delivery of over 300 interventions in year 2, supporting an estimated 262,000 children, young people and adults.

The effectiveness of individual commissioning processes and interventions could not be assessed at this stage due to recognising commissioning cycles and interventions being ongoing, limited progress in monitoring/evaluation and COVID-19 restrictions affecting delivery. However, it was evident that commissioning had evolved to become more strategic and to some extent more evidence-based across VRUs.

8. Conclusions

This final chapter presents conclusions from the year 2 VRUs evaluation on the overall extent of VRUs’ contribution to reducing, and preventing, SV. A brief consideration of further areas for research and investigation in future years of VRU operation is provided at the end of the chapter.

8.1 Overall contribution of VRUs

Overall, the evaluation evidence indicates VRUs have continued to make progress in implementing, and delivering on the aims of, a whole-system approach to violence reduction in their second year of operation. The contribution of VRUs was most evident in multi-agency working and data sharing. Some challenges, including those brought about by the COVID-19 pandemic, affected progress with elements of delivery.

In some aspects of their work, the evidence indicates that VRUs have variously acted as a catalyst for speed of change (i.e. some aspects of whole-system approach would have occurred but not within the timescales achieved by the VRU), for the substance of changes being made to prevent SV across the area (with VRUs’ strategic role in co-ordinating, providing extra resource to do so) and for both in some circumstances. Also, the evidence varies in its strength to demonstrate the extent of contribution towards different outcomes, which is to be expected from the longer-term nature of a whole-system approach, and VRUs varied in their starting positions prior to the funding.

There is strong evidence in this second year of their operation that VRUs have contributed to some of the specific, anticipated outcomes documented in the programme-level ToC. This contribution is observable to date in three overarching main ways and mostly in relation to the following outcomes:

• there is emerging evidence of SV funding reducing violence, and VRUs appear to be contributing to this; although reductions were concentrated on violence without injury offences, this aligns to the preventative nature of VRUs; based on the estimated number of offences avoided because of SV funding, there is an estimated benefit of £385m which represents a return on investment of £3.16 for every £1 of funding; VRU stakeholders were hesitant to claim attribution of impacts on SV, owing to only being in their second year of a longer journey, and the impact of COVID-19 on direct delivery, but are confident the VRU is the right response

• ‘partners/professionals buy into, and support, a whole-system approach’ and ‘Partners/professionals understand how their role contributes/aligns to Response Strategies / local action plans’; most VRUs have consolidated and strengthened their whole-system approach through: greater collaborative working with partners in year 2; broadening representation at the VRU’s strategic/governance levels to address some of the previous gaps from year 1; and introducing processes to enable a more balanced range of inputs to the work of the VRU, including those from non-statutory partners; this has created a more inclusive approach to the joint assessment, and prioritisation, of SV issues across the VRUs, specifically through the development of SNAs and action plans by joint working groups or committees set up to focus on specific VRU workstreams

• ‘SNA enables a focused and whole-system Response Strategy’ and ‘Partners/professionals understand and are better able to respond to individual/local needs’; across most VRUs, and multiple groups of stakeholders, there is clear evidence of a more widely held recognition of how valuable effective data and intelligence sharing is to producing an SNA that can profile the patterns of SV at an area-wide level while enabling a response to specific issues experienced by different localities within the ‘footprint’ of the VRU; in year 2, there is a wider awareness among partners/professionals of the need to build and extend on the knowledge and evidence available to the VRU, so as to better assess overall progress and support more informed decision-making and selection of interventions and engagement activity; there is also a greater appreciation, and more mature understanding, amongst VRU stakeholders that a sustained and continued commitment to data access and sharing is required to build on the achievements to date – manifested in, for example, VRUs’ intention/determination to make more progress in overcoming some of the challenges VRUs still face in accessing some local or sector-specific datasets

There is some limited evidence of VRUs beginning to make some contributions towards the outcomes related to reducing SV in the following two areas:

• ‘positive engagement with the VRU and partners’. VRUs’ year 2 plans for activities to engage young people and communities were hampered by several challenges, not least the impact that social distancing restrictions (as a result of COVID-19) had on limiting the face-to-face nature of activities, considered by many stakeholders to be the most effective forms of engagement work; however, there is evidence, albeit limited, that VRUs have delivered against their original plans to some extent and have learnt lessons from their implementation of some activities adapted for the pandemic; where they have been able to engage young people and communities, VRUs have set the foundations for year 3 with early signs of some positive outcomes for those groups who could be engaged. there is strong commitment to prioritise this work in year 3, based on an acute recognition of their delivery constraints this year

• ‘successful identification (referrals) and engagement of those at risk’ and ‘Improved perceptions of interventions/partners’. VRUs’ made good progress in delivering planned interventions; the young people’s voice captured through the evaluation evidence points to a consistently positive view of their engagement in, and experience of, VRU-supported interventions; there are also some early indications of positive impact on young people’s behaviours, mental health and wellbeing, which corroborates some of the available (albeit limited) evidence confirming the effectiveness of the nature and format of interventions commissioned and supported by VRUs

8.2 Key enablers

Enablers that were common across different elements of the whole-system approach and/or prominent included VRUs:

• being perceived as credible, independent (from the police) and non-duplicative, which supports engagement from partners and the sharing of data
• enabling other organisations to take the lead on specific work streams / actions to embed shared accountability
• harnessing multi-agency arrangements (and existing DSAs) to facilitate data sharing, and existing links to communities and young people
• sharing outputs (including the SNAs and Response Strategies) with partners and wider audiences to increase the profile of the VRU and demonstrate its value
• ensuring representation of communities and young people on strategic and operational boards to allow for feedback on the VRU approach
• having dedicated commissioning/intervention leads within the core team to oversee and align activity
• working closely with Surge to ensure preventative activity is reaching known hotspots

8.3 Future development of VRUs

In terms of the future development of VRUs, the evaluation findings support the following recommendations:

For the Home Office to:

• support VRUs to improve the quality of intervention monitoring data which informs, and underpins, evidence-based commissioning; while VRUs are becoming more evidence-based (e.g. commissioning based on the SNAs), high quality and consistent invention monitoring data would help develop the commissioning process and evidence base further

• urge other central government departments, including Department of Health and Social Care (DHSC) and Department of Education (DfE), to improve data access at VRU level and ensure partners take responsibility for their role and involvement in the VRU; this reflects the challenges identified in accessing certain data (e.g. hospitals) by some VRUs, and ensuring the progress made by VRUs and partners at the strategic level is operationalised

And for VRUs to:

• ensure clear, evidence-based rationale for commissioned interventions, supported by the YEF toolkit; this includes ensuring a clear link to the identified needs (SNAs), and evidence on the effectiveness of the intervention (or similar) in meeting these needs; for new/innovative interventions commissioned, this includes supporting the evidence base through monitoring and evaluation

• increase focus on effective engagement of VCS at strategic level; this is in recognition of the perceived value (e.g. insights to what will work operationally) of engaging grassroots organisations reported by the VRUs that could do so

• consider the potential role of external experts to navigate common data-sharing challenges and increase data analysis and insight capacity; there were several examples of this proving an effective strategy

• share any good practice toolkits or resource on implementing whole-system approach with all VRUs through learning networks

8.4 Areas for future investigation

The theory-based approach – utilising qualitative and quantitative evidence, and centred around ToCs – complemented with QEDs, has been critical to evaluating VRU approaches and progress within their varying contexts. It is recommended this approach is carried forward.

Although covered to a degree in the year 2 evaluation, research with community and voluntary organisations was limited. This was in part due to COVID-19 restrictions but also the distance of VRUs from some of these organisations. Future evaluation could aim to ensure greater coverage of these organisations, which would help understand working relationships with VRUs and the impact this has.

Similarly, future evaluation activity should prioritise research with young people. Now VRUs have a better understanding of the at-risk cohorts and approaches to provide support, it is an opportune time to capture (representative) views of young people across a range of interventions to understand the impact of support. This would include research with sufficient samples within interventions to assess the effectiveness of different types of interventions.

Annexes

Annex 1. Local evaluation summary

VRU Name	Local Evaluations - Summary and status - April 2021
Avon & Somerset	Internal VRU performance monitoring within OPCC led ‘hub’.
	Cordis Bright (commissioned Nov 2020) for process evaluation (timescales unclear – running into 2021/22).
Bedfordshire	Steve Broom independent consultant commissioned for process evaluation, taking a network analysis approach.
Essex	The Violence and Vulnerability Unit (VVU) is not commissioning a VRU-wide evaluation but will instead commission and support the evaluation of several interventions/initiatives.
	Internal resource for an evaluation impact officer proposed for 2021/22.
	VVU is focusing on commissioning an evaluation of a pilot project to support vulnerable adults, through Essex County Council Adult Social Care, but the evaluation has not been tendered at this stage and timescales are uncertain at this point.
Greater Manchester	The VRU has not commissioned a full programme evaluation, but there is a group of six activities/interventions (centralised commission projects) targeted for an external evaluation by Manchester Metropolitan University.
	Additionally, the VRU will provide in-house support and evaluation for Community Safety Partnership (CSP) / locality-led interventions funded through the VRU.
	Centrally-developed logic model informed by the SNA. Report due March 2021.
Hampshire	Evaluations are at each local VRU level:
	* Hampshire VRU: Solent University interim report produced (observation, interviews, focus groups, online questionnaires). Full evaluation ‘expected imminently’
	* Isle of Wight: internal case-studies
	* Portsmouth: University of Birmingham on Trusted Adults Worker model
	* Southampton: joint evaluation of A&E navigator with hospital (by MSc students)
Kent	Internal evaluation
	Force Inspectorate (VRU process)
	Continuous Improvement Team (VRU impact)
	Counterfactual Impact Evaluation (CIE).
Lancashire	Liverpool John Moore’s University
	UCLAN intervention specific evaluations
	UEA is picking up the ‘Caring dads: Safer children’ programme
Leicestershire	Evaluation lead Leicester University.
	Process evaluation runs from July to December 2020 and Outcome evaluation between July and March 2021.
	Process and Outcome evaluation findings will be published in March 2021 ‘year 2 Evaluation Report’.
	Internal local evaluation by Lead for Evidence and Evaluation colleague.
London	Tavistock Institution: review of interventions.
	Ecorys: process evaluation, reporting in May 2021.
Merseyside	Public Health Institute at Liverpool John Moores University (LJMU) in collaboration with other Merseyside Universities (process and impact evaluation)
	Focus on:
	* the whole-system approach
	* only a selection of interventions
Northumbria	Joint local evaluation Final report expected May 2021 [commissioned organisation not named] local evaluation which will focus on VRU and Surge activity in the area.
	Contribution analysis which will utilise both quantitative and qualitative data.
Nottinghamshire	Evaluation focusing on the impact of four interventions.
	Measure the impact of the Nottingham City and Nottinghamshire Violence Reduction Unit (NNVRU) at an intervention level, expected outputs and outcomes.
South Wales	LJMU to conduct the 2020/21 evaluation; monitor and measure the impact of the Violence Prevention Unit (VPU) on violence prevention.
	Network analysis, concept mapping, qualitative research with a systems lens (interviews with VPU partners and stakeholders) and analysis of programme data.
	Evaluation of the Wales Violence Surveillance System (now called Violence Intelligence for Prevention [VIP] hub) – this will include a survey with partners and interviews.
Thames Valley	The Thames Valley VRU is using a mix of internal and external evaluation resources. External evaluations have not been commissioned yet. The VRU will focus on evaluating a small group of (less resource intensive) interventions being delivered.
	The VRU is currently relying on internal evaluation, focusing on key commissioned interventions (DIVERT, CIRV, Mentors in Violence Prevention programme) and commissioned youth engagement activities (evaluation of the community engagement programmes run by Intensive Engagement and Redeeming our Communities). Internal evaluations are based on many different methods depending on the intervention – at least two interventions are planned to be evaluated with a quasi-experimental method, or at least a pre-post analysis with the use of control groups, while others are planned to be assessed with the use of contribution analysis and theory-based evaluation methods. Timescales and access to emerging findings are yet to be confirmed
West Midlands	Evaluations of:
	* Project-based activity – impact evaluation (interviews with providers and project leads, data collection and analysis of standard monitoring templates). Process evaluation (of four interventions, narrative literature review and mixed-method data review, stakeholder interviews).
	* Place-based activity (qualitative ethnographic studies, depth-interviews observations and informal community conversations in three place-based projects)
	* Programme-level activity (mixed-method, ToC and evaluation framework)
West Yorkshire	The local evaluation of the West Yorkshire Violence Reduction Unit (WYVRU) has been commissioned to an external evaluator - Sheffield Hallam University – and is currently ongoing. The evaluation aims to understand:
	* the collaborative relationship between the WYVRU and the West Yorkshire and Harrogate Health and Care Partnership (WY&H HCP), their application of the public health approach, good practice and how to further develop the partnership
	* what works in relation to current interventions, key factors for effective implementation and further development, and how best to evaluate impact
	This includes three strands of work:
	* Strand 1: case study on partnership between VRU and Harrogate Health and Care partnership
	* Strand 2: Literature review
	* Strand 3: Process and impact evaluation (focus groups, cost-benefit analysis, document review)
	The evaluation started in November 2020 and is planned to conclude in March 2021.
South Yorkshire	The local evaluation has been commissioned externally to Sheffield Hallam University and will:
	* review the effectiveness of the VRU in preventing and reducing violent crime so far
	* support development of a monitoring and evaluation framework that can measure effectiveness moving forwards
	* conduct impact feasibility studies on key interventions, outlining what data must be collected moving forward to enable measurement of impact
	The university will also conduct an online survey in local communities to establish perceptions of the effectiveness of the VRU and will engage with communities who have had direct involvement with the VRU, as well as those that have not.
	Learnings will be shared through:
	* VRU meetings organised by the Home Office
	* local learning events (e.g. learning lunches, N8 knife crime conferences, etc)
	* online platforms like Basecamp (VRU sharing platform) and the Public Health England Serious Violence Prevention Network
	* 2020/21 Communication Plan sharing lessons across South Yorkshire
	The evaluation started in November 2020 and is planned to conclude in March 2021.
Sussex	Awaiting further detail from the VRU on evaluation specifics

Annex 2. VRU evaluation framework

The evaluation framework provides the high-level research questions, set out by theme/sub-theme, that the evaluation will seek to address and indicates the data source(s) for each. Subsequent research tools will align to the framework and explore the research questions in greater detail. Table A2.1 summarises the themes.

Table A2.1: Summary of themes and sub-themes used in the evaluation framework

Theme	Sub-themes
Models of working	VRU leadership: exploring the different leadership approaches and their perceived effectiveness, what makes an effective leader and how leadership affects the VRU meeting its aims.
	VRU membership: exploring engagement from VRU members, links between VRU membership and leadership, and how membership affects the VRU meeting its aims.
	VRU model: exploring the perceived effectiveness of different VRU models, including how the different layers of the VRU interact (i.e. the ‘golden thread’) and how the model affects the VRU meeting its aims.
VRUs in the wider context of serious violence (SV)	Context: exploring VRU alignment with national and local strategies/policies related to SV and historical trends in SV.
	The impact of VRUs: estimating the overall impact of SV funding and disaggregating the impacts of VRUs and Surge within the wider SV context (above).
Whole-system approach (sub-themes adapted from WHO PHA definition)	Data and intelligence generation/sharing: exploring what data are being collected/shared, how well data are being shared, how data are being analysed and used, and how data generation/sharing affects the VRU’s ability to meet its aims.
	Strategic Needs Assessment (SNA) – identifying the target population: exploring how the VRUs are defining SV and understanding the at-risk (target population) cohort, and how this understanding affects the VRU’s ability to meet its aims.
	Multi-agency working: exploring the extent to which, and how, partnerships are effective and how VRU membership supports this, the effects of co-location and alignment with Community Safety Partnerships (CSPs) on multi-agency working, and how multi-agency working affects the VRU’s ability to meet its aims.
	With and for communities (and young people): exploring how VRUs are engaging with communities and young people and how these groups perceive VRUs, and how an approach with, and for, communities and young people affects the VRU’s ability to meet its aims.
	Response to SV: exploring how the whole-system approach has influenced the VRU’s response to preventing/reducing SV.
	Note, this sub-theme is purposefully cross-cutting with others related to the whole-system approach.
	Interventions: exploring which interventions have been developed/commissioned and how they align with the SNA and wider whole-system approach, the perceptions of interventions, and the experiences, outcomes and impacts of interventions on young people.
Support	Exploring how, and to what effect, the VRUs are supporting partner organisations and each other, and the (non-financial) support provided by the Home Office to VRUs.
Sustainability	Exploring funding and wider considerations to embed a whole-system approach to violence reduction over the longer term.

The different data sources are:

• document review: including funding applications, SNA, Response Strategies, and other documentation/analysis provided by the Home Office and the VRUs
• VRU leads: consultations with the VRU leads at three points over the course of the evaluation
• wider stakeholders: consultations with a wide range of stakeholders including VRU members, partner agencies and local communities
• young people: consultations/focus groups with young people involved with VRU activity
• counterfactual impact evaluation (CIE): comparing VRUs to areas that did not receive funding to, as robustly as possible, attribute impact to the VRUs
• quantitative analysis: drawing on available quantitative data to assess VRU outcomes
• local evaluations: evidence (quantitative and qualitative) from local evaluations; it is anticipated this will focus more on specific interventions funded by the VRUs

It is important to note that data sources, additional to the ones indicated already (see Table A.2), may become relevant for some research questions as the evaluation progresses. This is particularly the case for local evaluations, where limited information is currently known about the scope of these, and quantitative analysis, where the VRUs might be able to provide additional data to support some of the research questions. It is recommended that the evaluation framework is revisited when more is known about these data sources.

The evaluation framework is detailed in Table A2.2. While there is crossover between the research questions relevant to the process and impact evaluation (i.e. understanding the processes of VRUs reveals the underlying mechanisms of impacts), questions relating principally to the impact evaluation are highlighted in light blue. Furthermore, to avoid duplication, research questions have been allocated to the most appropriate theme but it is recognised there is some crossover (i.e. some research questions are interchangeable between multiple themes).

Recognising some themes in the evaluation framework were (to at least some degree) covered in the 2019/20 process evaluation, a common line of questioning against the different themes will be progress made and/or substantial changes since the first year of funding. Additionally, themes not covered or (owing to the evaluation focus, progress made by VRUs or data availability) not covered in detail in the 2019/20 evaluation will be prioritised in the 2020/21 evaluation – priority themes are indicated with an asterisk.

Table A2.2: Evaluation Framework

Theme / Sub-theme	Research question	Doc review	VRU leads	Wider stakeholders	Young people	CIE	Quantitative analysis	Local evaluations
Models of working
VRU leadership*	Has the VRU leadership changed since year 1? If so, why?	X	X
	What leadership approaches have been adopted (e.g. co-led by different orgs. or led solely by a single org.)? Why have these approaches been adopted (e.g. driven by culture, need, resource)? What are the perceived benefits/disadvantages of the different approaches?	X	X	X
	How is the VRU leadership impacting on the operational and frontline delivery of VRUs? Does this vary across different leadership approaches?		X	X
	What are the hallmarks of effective leadership? E.g. the background and skills of VRU directors / co-leads		X	X
	To what extent is the VRU leadership enabling the VRU to achieve its strategic and operational aims? Does this vary across different leadership approaches?	X	X	X
VRU membership	Has the VRU membership changed since year 1? If so, why?	X	X
	Has there been any changes to VRU members’ engagement since year 1? If so, why? Has COVID-19 impacted on engagement?		X	X
	Is there a relationship between VRU membership and the leadership? If so, what is this?		X	X
	To what extent is the VRU membership enabling the VRU to achieve its strategic and operational aims? Are there specific members/roles that enable aims to be met?	X	X	X
	In the absence of VRUs, to what extent would the different organisations (represented in the VRU membership) have worked collaboratively?	X	X	X
VRU model*	Has the VRU model changed/evolved since year 1? If so, why?	X	X
	How are the different layers of the VRUs (typically, strategic, operational and delivery focused) working together? In particular, to what extent is the strategic vision and public health (whole systems) approach of VRU leaders / governance boards reaching and being operationalised at the frontline?	X	X	X
	How has COVID-19 affected VRU models?	X	X	X
	How agile are the different VRU models in responding to complex and evolving local needs (including COVID-19)?	X	X
	To what extent are the different VRU models proving an effective way of working? What have been the success/challenges of these ways of working?		X	X
	To what extent is the VRU model enabling the VRU to achieve its strategic and operational aims? Does this vary across different models?		X	X
VRUs in the wider context of serious violence
Context	To what extent and how are the VRUs and Surge strategically aligned? How has this evolved since year 1?	X	X	X
	How have VRUs and Surge worked together to align their aims and operations?	X	X	X
	To what extent and how are VRUs aligned with the Serious Violence Duty?	X	X	X
	To what extent and how are VRUs aligned with other statutory functions (e.g. child safeguarding) and/or local policies?	X	X	X
	What were the serious violence trends prior to the establishment of VRUs?	X				X	X
	How do the VRUs define impact and what outcome measures are being used to monitor this?	X	X
	What external factors (e.g. COVID-19) have/may impact on the serious violence outcomes of interest?	X	X	X		X	X
Impact of VRUs*	What is the collective impact of VRUs and Surge (all Serious Violence funding)?		X	X		X	X
	What is the impact of VRUs (excluding Surge activity)?		X	X		X	X
	What additional impacts, at the strategic, operational and frontline, can be attributed to the strategic alignment of VRUs and Surge?		X	X
Whole-system approach
Data and intelligence generation/ sharing*	What data are being collected and/or analysed by the VRUs, how, by whom and why? Does access to (or lack of) specific data (e.g. domestic abuse) influence the focus of the VRU?	X		X
	How well are agencies sharing data, what added value does this give? Does this allow for Strategic Needs Assessment / intelligence to be updated and performance/impact monitored?	X	X	X
	How are the VRUs analysing data? And how is the analysis being used (e.g. SNA/RS, evaluation, audit, prediction, risk assessment, resource allocation)?	X		X
	How are the VRUs monitoring their performance/impact (including local evaluation)? How will this inform the future direction of VRUs?	X	X	X
	How is data sharing and analysis affecting the VRUs ability to meet its aims/intended impacts? What are the enablers/challenges associated with successful data sharing and analysis?		X	X
	In the absence of VRU funding, would data have been shared and analysed to same degree?		X	X
Strategic Needs Assessment (SNA) - Identifying the target population*	How are VRUs collectively defining serious violence and how does this influence the identification of the at-risk cohort (i.e. target population)? Is this definition consistently applied across the different levels of the VRU?	X	X	X
	How are VRUs collectively defining and forming an understanding of the at-risk cohort? Is this definition consistently applied across the different levels of the VRU?	X	X	X
	What are the common characteristics or key differences between at-risk cohorts, between (and within) VRUs?	X	X	X			X
	To what extent is geographical location a factor in defining/identifying the at-risk cohort?	X	X	X			X
	What risk factors were identified? How were these identified (e.g. any causal links)? How do different risk factors interact?	X	X	X			X
	Do VRUs anticipate the at-risk cohort to change over time (e.g. because of COVID-19, changes in SV trends)? How are changes to the at-risk cohort identified?	X	X	X
	How is the intelligence provided in the Strategic Needs Assessments affecting the VRUs ability to meet its aims/intended impacts? What are the enablers/challenges associated with this factor?		X	X
	In the absence of VRU funding, to what extent would the local approach be focused on the same at-risk cohort?		X	X
Multi-agency working*	To what extent and how have the VRU been working with partner agencies (including the police (beyond Surge))? What are the enablers/barriers to effective partnerships?	X	X	X
	How has the VRU membership/model enabled partnerships? Does this differ for partners not represented on the VRU membership?	X	X	X
	To what extent and how has co-location of VRU teams affected multi-agency working? Both from a day-to-day operational perspective and higher-level strategic perspective?		X	X
	How is devolvement to and/or strategic alignment with CSPs operating?	X	X	X
	What is the relationship between VRU’s responsibility to prevent violence and statutory agency responsibilities? What challenges and opportunities does this present?	X	X	X
	What hard and soft powers do VRUs have and/or lack to compel (coordinated) violence prevention activity in statutory agencies?		X	X
	How are the VRUs perceived by statutory agencies?			X
	How is multi-agency working affecting the VRUs ability to meet its aims/intended impacts? What are the enablers/challenges associated with successful multi-agency working?		X	X
	In the absence of VRU funding, to what extent would the local approach be multi-agency? (i.e. what additionality does the VRU bring to multi-agency working beyond existing partnerships (e.g. CSPs))		X	X
With and for communities (and young people)*	To what extent and how are the VRUs engaging with young people and the wider community? What are the enablers/barriers to successful engagement?	X	X	X	X
	To what extent are young people and the wider community aware / understand the public health (whole systems) approach? Are there effective ways (e.g. VRU branding, community panels) that support this?		X	X	X
	Is the approach to young people and wider community engagement consistently understood and adopted across the different levels of VRUs?		X	X
	How are the VRUs perceived by young people and the wider community (who might have poor historical relationships/perceptions of statutory services)?			X	X
	How is an approach with and for communities affecting the VRUs ability to meet its aims/intended impacts? What are the enablers/challenges associated with successful community and young person involvement?		X	X	X
	In the absence of VRU funding, to what extent would the local approach be with and for young people and wider communities?		X	X	X
Response to serious violence	How do the VRUs aim to prevent violence?	X	X	X
	How are VRUs utilising their Strategic Needs Assessments and Response Strategies to tackle serious violence (strategic and interventions)?	X	X	X
	How has the definition of serious violence influenced the VRUs approach/delivery (strategic and interventions)?	X	X	X
	How has the identified at-risk cohort influenced the VRUs approach/delivery (strategic and interventions)?	X	X	X
	How has engagement with young people and the wider community influenced the VRUs approach/delivery (strategic and interventions)?	X	X	X	X
	How has multi-agency working influenced the VRUs approach/delivery (strategic and interventions)?	X	X	X
	Does the definition of serious violence affect the VRUs ability to tackle linked issues, which are present locally (e.g. domestic violence, alcohol related violence)?	X	X	X
	Over what timescales, particularly for more longer-term strategic activity, are impacts anticipated to materialise?	X	X	X
Interventions*
	What interventions have been commissioned/developed and why? How do these align with the SNA and wider whole-system approach (both in terms of focus/delivery and being evidence based)?	X	X	X				X
	What is young people’s overall experience of the interventions received and their views on it?			X	X
	How are interventions (funded/developed by the VRU) affecting the VRUs ability to meet its aims/intended impacts?		X	X				X
	In the absence of VRU funding, would similar interventions have been commissioned and/or scaled-up?		X	X				X
	What are the outcomes and impacts of interventions on young people supported?			X	X			X
Support
VRUs supporting partners	How have the VRUs been supporting, financial or otherwise, local partners to reduce serious violence? Is this support perceived as effective? What elements work well and could be improved? How is this being measured/evaluated? (including COVID-19 emergency funding)	X	X	X
VRUs supporting each other	How have the VRUs been supporting each other (e.g. through learning networks)? Is this support perceived as effective? What elements work well and could be improved?	X	X
The Home Office supporting VRUs	How have the Home Office been supporting (non-financial) the VRUs? Is this support perceived as effective? What elements work well and could be improved?	X	X
Sustainability
Match funding	Have the VRUs received match funding or are expecting this over the longer-term? What are the enablers/challenges associated with securing (match) funding?	X	X	X
Long-term plans*	How is an effective whole-system approach to preventing violence embedded and sustained? What are the funding and (organisational) cultural/behavioural requirements and obstacles to this?		X	X
	What timescales are reasonable for achieving an effective whole-system approach to preventing violence and what are reasonable intervals for evaluating its effectiveness?

Annex 3. Quasi-experimental impact evaluation – technical detail

Methodological approach

Outcomes of interest

Aligned to the Home Office key outcomes for VRUs and the programme-level ToC developed, the following outcomes of interest were identified and accessed for the impact evaluation:

• NHS held data on hospital admissions resulting from intentional injury caused by a sharp object for:
  • all ages
  • those under 25 years old
• Police recorded crime for the following offence groups:
  • violence with injury
  • violence without injury
  • possession of weapons offences
• Homicide Index data

The above outcomes facilitate a thorough exploration of the potential impact of Serious Violence (SV) funding on violence. While not strictly violence, possession of weapons offences were considered as secondary outcomes. Robbery was also considered for analysis as a secondary outcome but did not meet the required modelling assumptions (common support assumption – see next section). Additionally (and intentionally not listed above), sexual offences and vehicle offences were explored to identify any underlying trends in crime, outside of the VRUs / SV focus.

While all data sources provide insights into violence, it is important to note their potential limitations. In particular, there might be increased detection within police recorded crime data resulting from increased enforcement activity (e.g. additional use of Stop and Search as part of Surge). To address this issue, we have sought, where possible and where suitable comparison outcomes exist, to triangulate findings using a crime and hospital admission data. Furthermore, a combined VRU and Surge ToC was developed as part of the evaluation to consider the interactions of prevention (VRU) and enforcement (Surge) activity on outcomes. The combined ToC is presented in Annex 4.

Analytical approach

To meet the aims of the impact evaluation, multiple quasi-experimental designs (QEDs) were employed. SV funding was awarded to the 18 PFAs experiencing the highest levels of violence. As such, simple comparisons to areas that did not receive funding (which were experiencing less violence per capita) are not appropriate. When funding (i.e. ‘treatment’) is not allocated at random, QEDs are required to construct a suitable counterfactual (which is an estimation of what would have likely happened in the absence of treatment). QED approaches included:

• Synthetic control methods (SCMs); SCMs enable appropriate comparisons by creating a synthetic control group, which is a weighted average of potential comparator areas; comparator areas that are more similar to the treated (SV-funded) areas on pre-intervention outcome trends receive a heavier weighting than those that are less similar; as well as providing reliable impact estimates (O’Neill et al, 2020), (generalised) SCMs provide additional benefits for the evaluation of VRUs:

• overall/average impact estimates for the entire treatment period, and average/cumulative impact estimates for each individual treatment period (e.g. months) can be calculated; this latter is particularly useful for VRUs where it is anticipated impacts will be gradual and accumulate over time; furthermore, it is possible to visualise these impacts, which allows exploration of the (theorised) initial impact of Surge activity (from April 2019) and then additional impact of VRUs (September 2019, when most were operational)

• impacts can be estimated at a whole programme-level (i.e. all VRUs combined) and for each treated area (i.e. each SV-funded PFA); while greater focus is placed on the programme-level analysis, which is the overall impact of SV/VRU funding, analysis of each funded area provides additional insight

The synthetic control groups were constructed following the approach developed by Xu (2017). For each treated area (e.g. PFA or CSP) a synthetic control group was constructed from a pool of comparator areas (all non-VRU areas) based on how similar they were (pre-intervention) to the treated area on the specific outcome of interest under investigation (i.e. the outcome impacts are being estimated for). Comparator areas that were most similar to the treated area received a higher weighting in the synthetic control group. For each outcome examined, diagnostic plots were assessed to ensure there was sufficient overlap between treated and comparator areas for impact estimates to be considered reliable, this is often referred to as the “common support” assumption.

• Interrupted time series (ITS); to further investigate potential impacts over time, ITS was employed on key police recorded violence outcomes ^{[footnote 20]}; based on pre-intervention trends in the treated area(s), ITS forecasts a post-intervention trend (which serves as a counterfactual) and compares this to the actual (or ‘observed’) post-intervention trend; by comparing the slope of forecasted trend to that of the observed, impacts can be estimate; a typical use of ITS is to estimate the effect of legislation on an outcome, such as the impact of the Health Act 2006 on smoking-related illnesses (Sims et al, 2010)

Besides the SCM and ITS approaches, fixed-effects regression, which is a generalisation of the difference-in-differences estimator, was used for comparison. While fixed-effects regression shares some similarities with SCM (comparing treated areas to comparator areas), the approach relies on the parallel trends assumption. This assumes that in the absence of intervention, the treated areas would have followed the same trend as the comparator (non-SV) areas. While this assumption cannot be proved (it is impossible to observe the treated areas without intervention), it is often justified through the examination of pre-intervention trends (see additional outputs section) and including covariates (additional factors that help explain the variance between treated and comparator areas) in the regression and/or limiting the comparator areas included to those that have similar characteristics (including pre-intervention outcome and/or factors related to the outcome). Recognising the SCM approach weights comparator areas based on the similarity of their pre-treatment period trends to treated areas (thus, relaxing the parallel trends assumption) and has been shown to provide reliable and accurate impact estimates relative to other approaches, the findings presented centre on this approach (O’Neill et al, 2020).

The specific QED approaches for the outcomes of interest were determined by the unit of analysis (i.e. geographical areas) and frequency (monthly, quarterly, annually) of available data. For each outcome of interest, Table A3.3: Outcomes of interest, data format, and analysis undertaken summaries the analysis undertaken at a programme level, and if estimating the impact of VRUs on SV independent of Surge activity was possible/appropriate.

Table A3.3: Outcomes of interest, data format, and analysis undertaken

Source/ outcome	Unit of analysis	Frequency	Programme-level analysis	SV / VRU impact
Hospital admissions	Local authority	Annual	SCM	SV estimates only
	Police force area	Monthly	SCM	SV and VRU estimates
Police recorded crime	Community Safety Partnership	Monthly	SCM	SV and VRU estimates
	LSOA	Monthly	SCM	SV estimates only
	Police force area	Monthly	ITS	SV and VRU estimates
Homicides Index	Police force area	Quarterly	SCM	SV estimates only

Recognising allocation to treatment was at a PFA level, where possible and appropriate, analysis focuses on sub-areas within PFAs. This is because at a PFA level, potential comparator areas, by definition, experience relatively less violence. Whereas at the LA/CSP level, there are areas within comparator PFAs that experience similar levels of violence (e.g. large towns and cities) to those within SV-funded PFAs. This is referred to as the common support assumption, which requires sufficient overlap between the treated and comparator areas.

Adjustment for impact of COVID

Analysis considered the impact of COVID-19 restrictions (national/local lockdowns). Specifically, it was theorised that national lockdowns (and wider restrictions) affected people’s movement (e.g. they were spending more time at home) so this impacted on their exposure to / opportunity for violence, which typically occur outside of the home.

Exploratory analysis using Google Movement data, which details the percentage change in movement patterns from a baseline of zero in January 2020 (see Figure A3.2), demonstrated people were spending more time at home (particularly during the national lockdown (April to May)), which coincided with a reduction in violence (see Figure A3.2). Regression analysis also revealed a statistically significant negative relationship between violence and residential movement.

With the above in mind, Google Movement data (aggregated at the local authority level and by month) was tested in exploratory SCM models as a covariate. Including movement data as covariate made only small differences to the SCM impact estimates. This was to be expected given VRU and non-VRU areas were similar in terms of average changes in movement (see Figure A3.2), and thus impacts on violence because of COVID-19 restrictions ‘cancelled out’. As such, and recognising the limitations of Google Movement data (in particular, data are not available before 2020 so natural changes, for example, people potentially spending less time at home during summer months, are not accounted for), the main results presented in the report do not include the residential movement covariate.

For the ITS models, a binary indicator (or ‘dummy’ variable) was included in the models for the national lockdown to ensure the post-intervention trend was not influenced by reductions in violence, which are attributed to the lockdown (rather than SV funding / VRUs).

Additional outputs

Figure A3.1 shows the monthly trends in all police recorded offences that were available/requested for the evaluation from April 2015 to September 2020. Data were at the Community Safety Partnership (CSP) level, which are broadly equivalent to local authorities. For violence with injury and violence without injury offences (the outcomes of interest for VRUs), the averages for CSPs in VRUs and CSPs not in VRUs followed a similar trend. Importantly, areas followed a similar trend over the COVID-19 period (a decrease during the lockdown are a return to pre-COVID levels in the summer months), which, for the impact evaluation, supports the argument VRU and non-VRU areas were comparable over this period.

Figure A3.1: Trends in police recorded offences, pre-SV funding, post-SV funding and during COVID-19

Adjusting for COVID-19 restrictions

Figure A3.2 shows the trends in Google Movement data between January 2020 and September 2020 for local authorities in VRUs and local authorities not in VRUs. The precise movement measure is percentage change in residential movement (i.e. time spent at home), which increased during the national lockdown in April, from the baseline of zero in January/February. We used this data as a covariate in exploratory SCM models to control for the impact of COVID-19.

Figure A3.2: Trends (and 95% CIs) in Google Movement data (2020)

Figure A3.3 shows the trends in Stop and Search rates for CSPs in VRUs and CSPs not in VRUs. Stop and search rates were rising in VRU areas prior to SV funding and continued to do so during SV funding. We included this data as a covariate in SCM models where we sought to isolate the impact of VRUs, as far as possible, from Surge activity (specifically, increased enforcement activity).

Figure A3.3: Trends in Stop and Search rates

Figure A3.4 shows the SCM for hospital admissions resulting from intentional injury from a sharp object for all ages. There was no statistically significant impact.

Figure A3.4: Hospital admissions resulting from deliberate injury from a sharp object (all ages) synthetic control group

Table A3.4 details the estimates for the ITS models conducted at the programme level. Figure A3.5 and Figure A3.6 present the results for the models where the intervention period was set at September 2019 (April 2019 models are presented in the main report).

Table A3.4: ITS programme-level estimates (asterisks indicate statistical significance (95%))

	Violence with injury (Sep 2019)	Violence without injury (Sep 2019)	Violence with injury (Apr 2019)	Violence without injury (Apr 2019)
Coefficients	# Estimates
Step-change	-715.77	119.21	-388.96	1531.72*
Slope	-258.87*	-132.18	-228.23*	-164.66
Lockdown	-2559.72*	-1800.69*	-2601.34*	-1595.15*

Figure A3.5: ITS of police recorded violence with injury (all SV-funded areas) – September 2019 (VRUs operational) intervention start

Figure A3.6: ITS of police recorded violence without injury (all SV-funded areas) – September 2019 (VRUs operational) intervention start

Annex 4. Combined VRU and Surge ToC

Annex 5. Examples of community and young people engagement

Table A5.5: Examples of community and young people engagement

Campaigns and communications

One VRU developed a campaign which aimed to raise awareness of county lines and cuckooing among parents, the general public and young people. It developed a range of targeted messages for each of the audiences using feedback and messages from communities and young people. This campaign has now been adapted by two other VRUs across the country.

“We got young people from alternative provisions, so very much at-risk youth who were very much engaged with agencies or community organisations, and sat them in a room and said, ‘What would literally stop you in your tracks? What would change your behaviour? What would make you stop and think?’ And from that we got some quite raw audio and footage of those young people talking about their experiences with serious violence and growing up specifically in [VRU Area] and we’ve been able to pull out this amazing social media campaign over a number of weeks and the reach of that has been absolutely phenomenal.” VRU lead

Another VRU worked with the police and used its social media accounts to alert local communities to the times in which police officers were due to be present in their local area. The police would tweet to let communities know where they would be on certain days and times. The feedback suggested that local communities appreciated this approach as they had, at times, asked the police to check specific areas where they felt unsafe.

“One thing they’ve done, and the local councillors praised it the other day, was massively improve their presence on social media. They’re saying, ‘We’re out here, Friday night, we’re the police of your community, direct us to any local hotspots.’ The community are saying ‘Can you check this street, that street?’ They turn up there and they’re showing it on social media saying, ‘All quiet here, we’re dealing with a situation, we’re done here.’ The community are saying thanks for that. This is a community that said we don’t see the police, that’s helped.” External partner

Engagement events and consultations

One VRU ran a series of workshops with local schools about SV. Prior to the workshops, it also engaged with young people to help them design the materials, agenda and key messages to make sure that the content was relatable.

Another VRU trialled a new digital tool for community and young people engagement. It used the tool to enable young people and communities to feedback on specific issues such as what worries them most or what works well in helping to tackle SV. The results were analysed and presented back to senior stakeholders and used to inform a community action plan.

Dedicated feedback groups or mechanisms

One VRU created a Young Leaders Action group to help directly influence its strategy and plans. The VRU employed the young people to ensure the youth voice was embedded in their decision-making process. They were involved in the commissioning process and influenced which interventions were funded. The young people involved noted that it had given them ‘a voice’ and welcomed the opportunity to contribute in what they perceived to be a new and positive way.

Another VRU had run several community engagement events throughout 2020 and established three Community Action Groups as a result. The action groups had identified and agreed on three priority areas of focus for the locality: mental health and wellbeing, youth narrative and youth justice. These action groups were feeding into the local Problem-Solving Partnership / task force which drives the VRU local activity.

“The thing is that it’s very important. It’s important for young people to get their opinions across and tell people how they really feel. Like, the more young people that do that, then the more people know what they could do. They need to listen to young people and that is what I want and to see things happen because they listened.” Young person participating in community youth group

Getting feedback through interventions, co-production and co-design

Several VRUs were collecting feedback through their commissioned interventions. They usually facilitated this through pre and post questionnaires or through informal conversations with the frontline staff asking about changes in wellbeing, attitudes and behaviour, and perceptions of the interventions themselves. A few delivery organisations then shared this informal feedback with the VRUs through their monthly and quarterly reports.

One VRU introduced several community-led pilots designed by Community Service Volunteer (CSV) organisations. The programme placed community co-design and their priorities at the heart of the interventions by providing community organisations with targeted funding. In practice, the pilot involved a community lead facilitating a session with community members to discuss their views and perceptions of community safety and the challenges faced in their local area. The feedback was used to produce priorities for each area, which were translated into a commissioning brief. Community representatives were then involved in the commissioning process and feedback and evidence was collected through evaluation and monitoring.

“Well, the Serious Violence Action Plan, and the priorities and the commitments were shaped by that community work. So, it’s based on what communities were telling us. Like the school-based police officer policy that came about because we engaged every school in [area] on violence. So, that document, which is effectively our Bible, was largely shaped by us reading really, data, communities and best practice, nationally and internationally. So, we used those 3 things together to create that action plan, so that’s how communities influence the programme as a whole.” External VRU partner

References

Foster, R. (2013) ‘Knife crime interventions: what works?’ The Scottish Centre for Crime and Justice, Research Report No. 04/2013.

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 (2019) ‘New public health duty to tackle serious violence’.

Home Office (2020) Violence Reduction Unit Interim Guidance.

NPC (2020) ‘Trauma-informed approaches: what they are and how to introduce them’.

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

Public Health England (2019) A whole-system multi-agency approach to serious violence prevention: A resource for local system leaders in England.

Robbins, M.W. & Davenport, S. (2021) ‘microsynth: Synthetic Control Methods for Disaggregated and Micro-Level Data in R’. Journal of Statistical Software, vol. 97(2), pp. 1-31.

Sims, M., Maxwell, R., Bauld, L. & Gilmore, A. (2010) ‘Short-term impact of smoke-free legislation in England: retrospective analysis of hospital admission for myocardial infection’. British Medical Journal, vol. 340:c2161.

Warren F. (2016) ‘‘What Works’ in Drug Education and Prevention?’ Scottish Government Social Research.

Xu, Y. (2017) ‘Generalized Synthetic Control Method: Causal Inference with Interactive Fixed Effects Models’. Political Analysis, vol. 25(1), pp. 57-76.

Footnotes

1. The return-on-investment figures are based on the central estimate of violent offences prevented. Recognising a degree of uncertainty around the central estimate, the estimated return-on-investment range (in which there is a 95% probability the ‘true’ return on investment falls) was between £0.40 and £11.03. ↩

2. To enable impacts to be estimated, data for the outcomes of interest is required before and after SV funding was introduced and, for the SCMs, covering VRU and non-VRU areas. ↩

3. The SCMs used a fixed-effects estimator to control for differences between areas (e.g. in outcomes recording, unobservable/time-constant factors) ↩

4. SCMs are less susceptible to regression to the mean, and better capture the “additionality” of SV activity over other national/local co-interventions. ↩

5. Additionally, autocorrelation and seasonality were adjusted for (fixed-effects estimator for SCMs and ARIMA models for ITSA). ↩

6. 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. ↩

7. Recognising there was no statistically significant impact on possession of weapons offences, cumulative impacts are not depicted. ↩

8. PFAs not included in the analysis were Humberside, Sussex, Kent, Wiltshire (alongside Greater Manchester and Wales where offence data was unavailable) ↩

9. A slightly different SCM, designed specifically for smaller geographical areas, was applied. See Robbins, M., & Davenport, S. (2021). ↩

10. Only PFAs with high and consistent levels of geocoding were included in the analysis. ↩

11. Recognising Google movement data is only available at Local Authority- / CSP- level, analysis was restricted to up to March 2020 (pre-COVID-19). Furthermore, the outcomes were not population adjusted (i.e. to rates) as LSOAs are of a similar size. ↩

12. Following the approach of Heeks et.al (2018). ↩

13. Uprated from estimates provided by Heeks et.al (2018). ↩

14. Year 1 funding + 50% of year 2 funding. Excludes Greater Manchester and South Wales VRUs, where offence data was not available. ↩

15. The 5 C’s are: collaboration with local stakeholders, co-production with partners and communities, co-operation in data and intelligence sharing, counter-narrative to create systems that mitigate against the risk of young people engaging in serious violence and community consensus, by working with communities to empower them. See: Public Health England (2019). ↩

16. While there is not an agreed definition of a ‘trauma-informed approach’, NPC outlines five principles for providing care in a trauma informed way: recognising and responding to trauma, providing safe environments, taking a strength-based view, building empowering relationships and promoting equality of access (NPC, 2020). ↩

17. This figure is based on monitoring data from 17 of the 18 VRUs. ↩

18. 12 young people that had taken part in interventions were interviewed. ↩

19. Due to limitations experienced as a result of the COVID-19 restrictions, the young people element of the evaluation adapted its approach to gain a breadth of coverage across several interventions, as opposed to focusing the interviews to collect concentrated data from a limited number of interventions. ↩

20. ITS is less appropriate for outcomes with small counts / volatility such as homicide. ↩