Authors: Ellie England and Patrick Corcoran, in partnership with Frazer-Nash Consultancy and Crest Advisory

Executive summary

Background

In January 2023, the Home Office (HO) announced the national expansion of Clear, Hold, Build (CHB) to every police force in England and Wales by March 2024, to improve the response to serious and organised crime (SOC) at the local level. The core aims of CHB are to reduce the SOC threat, reduce crime, and help local communities become more resilient to SOC in the future.

CHB is a place-based operational framework which provides a model for uniting the resources of the police, partner agencies and the local community. CHB is designed to be cost efficient through utilising existing resources and assets in an area. The framework comprises 3 phases, designed to be deployed flexibly:

• Clear removes the immediate SOC threats through co-ordinated enforcement work by police and partners to target and disrupt organised crime groups (OCGs) and their spheres of influence through arrests and disruption activity
• Hold involves interventions and further disruption activities to consolidate and stabilise the Clear phase, so that remaining OCGs cannot fill the void created
• Build involves delivering a whole-system approach through interventions that tackle the drivers of crime and reduce the risk of SOC re-emerging in the future

Annex G shows the CHB programme-level theory of change (ToC), which sets out the intended outcomes and impacts of CHB, and the causal pathways for achieving these through each phase.

Aims and methodology

The evaluation included, to some extent, 13 pilot sites across 10 police forces. It aimed to understand the effect of CHB on crime and community perceptions and resilience, and to understand the factors affecting implementation. To address these aims, the HO commissioned an independent impact evaluation by Frazer-Nash Consultancy and Crest Advisory, while HO analysts undertook the process evaluation. The impact evaluation had 2 key workstreams:

• crime analysis: aggregate-level impacts on 5 crime groupings were estimated using a combined synthetic controls and difference in differences approach – these crime groupings were: total police recorded crime (PRC); antisocial behaviour (ASB) proxy (arson, criminal damage, and public order offences); acquisitive crime (burglary (residential and business), robbery, and theft of a vehicle); drugs (possession and trafficking of drugs); and violence (violence with and without injury)
• community survey: impact on community perceptions of crime, safety, and resilience was estimated by conducting and analysing a retrospective survey

The process evaluation used a range of methods to understand the factors affecting implementation. These were: a proforma survey, in-depth interviews with police and delivery partners from 6 pilot sites and an online stakeholder survey.

Key findings from the impact evaluation

Across all pilot sites in total (at aggregate level), it is estimated that CHB resulted in a statistically significant reduction in the acquisitive crime grouping. The effect size was equivalent to a 23.7% reduction in acquisitive crime, or 51 crimes prevented per month across CHB areas, compared to the control group^{[footnote 1]}. No statistically significant impacts on the other crime groupings were found at aggregate level. However, sensitivity analysis and further analysis found that CHB can be effective at reducing other crime groupings. Statistically significant reductions in all 5 crime groupings were found for the following analyses:

• when the 2 worst performing pilots were removed from the sample
• in sites that were identified through a data-driven approach (using harm and OCG mapping)
• in medium-sized sites (approximately 1 to 3km²⁾

The results indicate substantial potential for CHB to reduce a range of crime types when applied in the right conditions.

Results from the community survey were less positive, possibly due to several methodological limitations. In particular, the survey had a low response rate (less than 2%), increasing the risk of non-response bias. It may be the case that the survey respondents hold more extreme views than the sample that did not respond, reducing confidence in the representativeness of the survey findings. Furthermore, it is possible that the increased enforcement activity may have had unintended negative consequences in pilot sites by drawing attention to crime and wider issues. These limitations should be considered when interpreting the results.

Across most metrics, the survey found no statistically significant differences in the level of change reported by respondents in the pilot versus control areas. However, in some cases, there was evidence of negative impacts. Compared to control area respondents, respondents in pilot areas were significantly more likely to say that things had got worse for a subset of metrics under community resilience and cohesion, the local environment, and perceptions of crime and policing.

Key findings from the process evaluation

The process evaluation found that sites valued CHB for the framework it provides in embedding a multi-agency and ‘4P’ approach (the Pursue, Prepare, Protect, Prevent principles originally developed to tackle terrorism) to reducing SOC and its harms. There was strong consensus that CHB has improved partnership working, creating a shift from siloed to collaborative working practices between police and delivery partners. Key delivery partners included the local authority (particularly staff working in the housing, environmental, or youth services departments), trading standards, and community organisations.

HO practitioner guidance sets out several operational elements for effective implementation of CHB. Generally, sites seemed to follow these recommendations. For example, most sites adopted a data-driven approach to selecting their CHB area and had a multi-tier governance structure in place.

Sites adopted a collaborative approach to implementation. During the design stage, police forces utilised support from key partner agencies to collate information on the key organisations and interventions in the CHB area (called an asset map) and identify gaps in local services. Sites also consulted the community to help select interventions to be delivered through their CHB programme, facilitating engagement.

During the delivery stage, police and delivery partners were involved in each phase, though Clear was mostly led by the police and Build by partners. Phase sequencing was flexible, with phases often running concurrently. In addition, a range of interventions were delivered as part of each phase, including undercover operations and days or weeks of action for Clear; hotspot policing and ASB patrols, environmental impact days and community engagement activities (such as community hubs and meetings) for Hold; and educational programmes with parents and young people on criminal exploitation, employment schemes with local businesses and youth-focused activities (such as creating youth centres) for Build.

The process evaluation identified a set of key facilitators to implementation, including:

• effective partnership working, achieved through establishing co-ownership of the programme, shared understanding of CHB and SOC, and a clear governance structure
• community involvement, by consulting the community early in the design stage and using a range of mechanisms to update the community on progress
• adopting an intelligence and data-driven approach to programme design
• senior support, particularly for engaging delivery partners and securing resources

Although there were no critical challenges, key barriers identified included:

• resourcing issues, particularly the lack of bespoke funding and capacity constraints, which sometimes undermined the quality of implementation
• low community engagement and lack of buy-in from some delivery partners created challenges in the Build phase and disrupted momentum

Overall learnings and recommendations

A full list of recommendations is in section 6.3. Key learnings for the future implementation of CHB drawn from across the impact and process evaluation include:

• using data to inform programme design is key to ensuring the most appropriate hotspot is selected, the activities chosen can address the target issues, and there is shared awareness of the key problems in the community
• co-ownership between police and partners should be established early on; involving partners in the programme design and educating partners on their relevance to SOC is key to achieving this
• a multi-tier governance structure should be established to help keep the project on track by providing accountability and a clear escalation structure for issues to be resolved swiftly
• consulting the community early on in the design stage ensures that the programme is designed to address their concerns, maximising engagement
• the size of a hotspot is important and requires consideration – an area needs to have sufficient local assets to leverage to develop a comprehensive Build phase, and an area should not be too large as this dilutes resources
• delivery should be flexible – phases can run concurrently and there is value in starting with Build
• the government should consider providing bespoke funding for CHB – the eligibility criteria for any future bespoke funding would need to be carefully considered to overcome common challenges associated with other funding streams (such as the short-term nature of funding) and prevent undermining the sustainability of CHB (which is a risk if funding were to be removed)

Recommendations for future evaluations include:

• alternative approaches to gathering community perceptions should be explored to achieve a better response rate; for example, face-to-face surveys
• an economic evaluation through a robust cost-benefit analysis should be undertaken, to understand whether CHB is value for money

Acknowledgements

First, the authors would like to thank the research teams at Frazer-Nash Consultancy and Crest Advisory. In particular Lewis Baxter, Sam White, Michael Hall and Sarah Hutton at Frazer-Nash Consultancy, and Oli Hutt, Serina Fuller, Alex Stacey and Fernanda Reynoso-Serna at Crest Advisory. Thanks also to colleagues at the Home Office who supported with fieldwork and analysis: Abbie Harrison, Alex Harris, Ellie Daly, Issabel Faithfull, Izzy Dunlop, James Mulcahy, Matt Roberts, Mollie Nielsen and Olivia Jeffery. Thanks to the police forces who facilitated the evaluation and gave their time to the research. Lastly, thanks to the two peer reviewers who reviewed and provided comments on an earlier draft of this report.

1. Introduction

1.1 Policy background

In January 2023, the Home Office (HO) announced the national expansion of Clear, Hold, Build (CHB) to every police force in England and Wales by March 2024, to improve the response to serious and organised crime (SOC) at a local level. Prior to this, 7 police forces had been piloting the approach since 2020. Several fatal shootings involving innocent members of the public relating to rivalry between organised crime groups (OCGs) in Merseyside in 2022 accelerated expansion plans, leading to the announcement of a national roll-out.

Defining SOC is complex and there is no widely accepted definition (Kirby et al., 2018; Lavorgna et al., 2013; Pitts et al., 2017). Despite this, the HO defines SOC as ‘individuals planning, co-ordinating and committing serious offences, whether individually, in groups and/or as part of transnational networks’ (Home Office, 2023a). Examples of serious offences covered by the term are child sexual exploitation and abuse; illegal drugs; illegal firearms; money laundering and other economic crime; and modern slavery and human trafficking. SOC is estimated to cost the UK at least £47 billion annually, although this is likely to be an underestimate (Home Office, 2023a).

Much of the impact of some types of SOC is felt in local communities where OCGs operate. At a local level, SOC often manifests as illicit markets linked to drugs, stolen or counterfeit goods, and labour exploitation (Crocker et al., 2017). These activities have economic and social impacts on the local community, although impacts differ by local conditions, context, and type of market. For example, some UK studies have found evidence of OCGs using violence and intimidation to incite fear among local residents and businesses, preventing them from reporting intelligence to the police (Crocker et al., 2017; Crocker et al., 2019., Lupton et al., 2002; Campana & Varese, 2018). In addition, OCGs have been found to coerce and exploit vulnerable members of communities, particularly the youth, to carry out drug dealing activities (Hales and Hobbs, 2010; Crocker et al., 2019). However, other studies have found that violence is rare in some illicit markets (Coomber, 2015). There is also evidence to suggest that local communities can be complicit in the success of illicit markets, benefitting financially due to, for example, the economic opportunities provided for people whose formal labour prospects are weak (Lupton et al., 2002). Building the resilience of local communities against organised criminal activity is a key priority set out in the UK government’s current SOC strategy (published under the 2022 to 2024 Conservative government).

To address SOC at a local level, CHB promotes a whole system-approach by providing a framework for police forces, supported by Regional Organised Crime Units (ROCUs)^{[footnote 2]}, to work closely with partners from local authorities and other public sector organisations, the voluntary and charity and private sectors in reducing SOC and its harms. The number of sites has grown substantially since the announcement of national expansion. In July 2024, CHB was operational in 40 of the 43 territorial police forces with a total of 59 sites across England and Wales.

1.2 Overview of the CHB approach

CHB was developed by the HO to reduce the impact of SOC in locations of highest harm, drawing from the principles of the military ‘Clear, Hold, Build’ counter-insurgency operational approach adopted by US and British army forces in Iraq and Afghanistan^{[footnote 3]}.

In the SOC context, CHB is a place-based framework that provides a model for uniting the resources of the police, partner agencies and the local community, integrating the 4P framework for reducing the level of SOC (Pursue, Prevent, Protect, Prepare). As an operational framework, CHB is designed to be cost efficient through utilising existing resources and assets in an area. It is therefore not supported by bespoke funding from the HO, except in the case of Merseyside which received £350k in 2023 to facilitate their expansion of CHB.

The framework comprises 3 phases, designed to be deployed flexibly. Although these phases can happen in succession, they can also be implemented simultaneously and should be repeated as necessary. A brief summary of each phase is described below; however, see section 5.1.4 for a more in-depth overview.

• Clear: involves co-ordinated enforcement work by police and partners to target and disrupt OCGs, their members and spheres of influence through arrests and disruption activity; this phase aims to clear the area of immediate SOC threats
• Hold: involves interventions and further disruption activities to consolidate and stabilise the Clear phase, which can create a power vacuum; this phase aims to hold the area so that the remaining OCGs or members cannot fill the void created
• Build: involves delivering interventions with the community that tackle the drivers of crime and reduce the risk of SOC re-emerging in the future; this phase aims to build community resilience and improve the working, living and recreational environment of an area so that the community is less susceptible to the draw of crime and OCGs

Through successful implementation, CHB aims to reduce the SOC threat, reduce crime, and help local communities become more resilient to SOC in the future.

2. Evaluation methodology

2.1 Evaluation aims

The 3 overarching aims of the evaluation were to:

• understand the effect of CHB on crime
• understand the effect of CHB on community perceptions and resilience
• understand the factors affecting implementation, to improve future delivery

To address these aims, an evaluation comprising 2 overarching elements was conducted: an impact evaluation and a process evaluation. The HO commissioned an independent impact evaluation by Frazer-Nash Consultancy and Crest Advisory, while HO analysts undertook the process evaluation. Key research questions the evaluation sought to answer include:

• what is the impact of CHB on crime levels?
• what is the impact of CHB on community perceptions of crime and environmental improvements?
• what is the impact of CHB on community resilience?
• to what extent does CHB improve collaboration between the police and other agencies in tackling SOC?
• what are the facilitators and barriers to successful delivery of interventions for each CHB phase?

There were 13 pilot sites across 10 police forces included, to some extent, in the evaluation. Broadly, the HO purposively selected these sites to ensure a range of representation was achieved in terms of the SOC threats targeted, geographic spread, and their level of maturity. Practical reasons were also considered. For example, initial engagement with forces found that some sites did not have the capacity to support the evaluation and its research activities, so these sites were deemed out of scope. The number of sites included in each element and rationale for selection is detailed in the following sections.

2.2 Impact evaluation methodology

This section outlines the impact evaluation approach and research methods. See Annex A for a technical methodology.

2.2.1 Impact on crime

Frazer-Nash Consultancy measured the impact of CHB on crime using police recorded crime (PRC) data obtained through the HO’s Data Hub (HODH). The analysis focused on measuring the impact at aggregate level to understand the average effect of CHB, with all 13 evaluation pilot sites included in the analysis. Five crime groupings were identified for the impact evaluation, based on the priority crime types (as indicated by the proforma survey) and/or those crime types that are more likely to be linked to SOC. The groupings were:

Total PRC: the sum of all individual crime types collected via the HODH.
Antisocial behaviour (ASB) proxy:^{[footnote 4]} comprising arson, criminal damage and public order offences. This is a proxy measure for ASB offences.
Acquisitive crime: formed of burglary (residential and business), robbery, and theft of a vehicle.
Violence: violence with or without injury.
Drugs: possession and trafficking of drugs offences.

Data on the number of knife-enabled offences was also collected from the HODH for both pilot and control areas. However, knife-enabled offences are not a crime type in themselves, but a flag used by police to record whether a knife was involved in an offence. As a result, this was treated separately to the 5 crime groupings in the analysis to avoid the double-counting of offences in impact estimates. Impact was estimated using a before and after comparison of the average knife-enabled offence rate in pilot areas versus control areas.

It is important to highlight that measuring the impact of CHB on the following types of SOC was deemed out of scope of this evaluation due to data limitations:

• child sexual exploitation and abuse (CSEA): the high levels of under-reporting of CSEA mean that PRC data do not provide a good estimate of prevalence
• modern slavery: the low volume of modern slavery offences due to the small geographic areas of CHB sites means detecting impact is extremely unlikely
• fraud: the HO does not have access to detailed fraud data as Action Fraud records these via a different system

For the 5 crime groupings, Frazer-Nash adopted a quasi-experimental method to estimate the impact of CHB using a combined synthetic control and difference in differences (DiD) approach. This essentially involved comparing the outcomes in the CHB areas (treatment areas) to outcomes in a group of statistically matched and weighted control areas (non-treatment areas). Broadly, a 4-step approach to analysis was taken:

1. Demographic data from the 2021 Census were collected for each pilot site. Each pilot site was matched to non-treatment areas with similar demographic characteristics within the same police force area (see Annex A for further details on the matching process and demographic variables used).
2. To improve the similarity of the control group (the matched non-treatment areas), a synthetic controls method was then employed. This weighted each non-treatment area according to how closely their pre-intervention crime trends matched the trends in the pilot area.
3. Verification checks were undertaken to validate the similarity of the synthetic control’s crime trends. These checks involved a visual inspection and a statistical test called an event study, which checked for any statistically significant differences in crime trends between the pilot site and synthetic control. This step confirmed whether the synthetic control provided a good estimate of what would have happened to crime in the pilot site, without CHB.
4. Data was aggregated across CHB pilots and the synthetic controls to form 2 all-area comparison groups. Impact was measured using DiD, which compared the pre-post change in crime rates for the pilot group to the pre-post change for the synthetic control group. CHB was considered to have had an impact if any difference found was statistically significant at the 95% confidence level. This indicates that we can be 95% confident that the result observed is not due to chance.

In addition to understanding the average impact of CHB on crime for the full sample, further analysis was conducted to understand where CHB works well and less well by grouping pilot sites by shared characteristics. The analysis covered 3 themes:

1. CHB launch (the reason for initiating CHB).
2. Delivery progress (the phases implemented so far).
3. Area type (the properties that define a pilot area).

More detail on the themes is provided in section 4.1.3.

Crime displacement analysis was also undertaken at pilot level. This was conducted by mapping a buffer zone around each pilot area and collecting crime data from the HODH for this zone. An approach called Bayesian time series (BTS) was used to measure crime displacement, which analyses differences between expected and actual post-CHB crime trends. The expected crime trend is modelled using previous crime trends and estimates what would have happened had CHB not been implemented. If the analysis found a statistically significant difference between the expected versus actual crime trends post-CHB in the buffer zone, this was seen as evidence of crime displacement. See Annex A for more detailed information on BTS.

2.2.2 Impact on community perceptions and resilience

Crest Advisory measured the impact of CHB on community perceptions and resilience using a quantitative survey, henceforth referred to as the ‘community survey’. The community survey included questions covering a range of outcome metrics, including community cohesion and resilience, perceptions of crime and safety, and perceptions of reporting crime and policing. Annex C contains a full overview of the question set.

Due to budgetary constraints, it was only feasible to include 6 pilot sites and their control areas (as identified through the analysis outlined in section 2.2.1) in this element of the impact evaluation. The pilot sites were selected purposively, based on those forces which had already given initial agreement to the HO to participate in the community survey. However, a secondary consideration arose during the evaluation design stage relating to site maturity. Each site included had to have been active for around 12 months. This was considered a sufficient period for any changes in perceptions to be achieved and detected, while also being a reasonable period for respondents to recall their pre-CHB perceptions, so that impact could be measured retrospectively.

With the exception of one site, a letter inviting residents to complete an online survey (known as a push-to-web survey) was sent to all households in pilot areas and a subset of control areas matched through the method described in section 2.2.1. A hard-copy version was also available to increase inclusivity. A reminder postcard was then sent 2 weeks before the survey closing date to help increase the response rate.

One site was excluded from the analysis due to the extremely low response rate achieved. This was to prevent respondents from that pilot site being over-represented in the data. The survey achieved a 2.0% response rate (385 completes) across the 5 pilot sites included in the analysis, and a 1.7% response rate (250 completes) across the control sites (see Annex A for additional detail on the response rate).

Due to the timing of the evaluation, it was not possible to conduct surveys both before and after CHB had been implemented. Instead, a single survey was conducted utilising a retrospective approach to measuring impact. This involved asking respondents how they felt about each factor on the day of the survey and then whether this factor had become better, worse or stayed the same compared to a year ago.

Following data collection and cleaning, response data was weighted to improve the representativeness of survey results (details on the weighting method are in Annex A).

Once data weighting was complete, a weighted t-test was used to test whether there were any statistically significant differences between the pilot and control samples in the proportion of respondents who reported experiencing positive or negative change. The report focuses on the results that were statistically significant at the 95% confidence level. This means that we can be 95% confident that the result observed is not due to chance.

2.3 Process evaluation methodology

The process evaluation aimed to understand the key factors affecting implementation. This research adopted a mixed-methods approach formed of 3 strands, each summarised below.

2.3.1 Proforma survey

A proforma survey was distributed to all 13 evaluation sites in September 2023. This collected important information on each site to inform the final evaluation design and improve the understanding of CHB programmes being delivered. The survey collected information on programme design, the contact details of key stakeholders and details of other crime interventions in the CHB area. This data was descriptively analysed.

2.3.2 In-depth interviews

Interviews with a range of policing stakeholders and key stakeholders from other agencies and organisations (referred to as ‘delivery partners’ henceforth) involved in implementing CHB in 6 evaluation sites were conducted between November 2023 and February 2024. Sites were selected for interviews to ensure a range of police forces and sites at varying levels of progress were represented.

The interviews aimed to generate a rich understanding of the facilitators and barriers experienced by stakeholders implementing CHB. There were 2 phases:

Phase 1 explored the set-up and design stage of CHB through a range of topics, including: the CHB team, community asset mapping, and engaging delivery partners; only policing stakeholders were interviewed for this phase.

Phase 2 focused on the delivery stage of CHB through a range of topics, including: delivery of each phase, community engagement, multi-agency working, and perceptions of impact; policing stakeholders and delivery partners were interviewed for this phase.

Table 1 shows that 43 interviews were conducted in total. Participants were selected using a combined purposive (using contact details supplied in the proforma) and snowballing sampling (using suggestions from forces) approach. Each interview was recorded, transcribed and coded. Codes were then reviewed, summarised and thematically analysed.

Table 1: Interview sample

	Police	Delivery partners	Total
Phase 1	13	0	13
Phase 2	13	17	30
Total	26	17	43

2.3.3 Online survey

An online survey was distributed to stakeholders involved in implementing CHB in the 13 pilot sites in March 2024. This survey aimed to generate a quantitative understanding of their experiences implementing CHB and perceptions of the framework. Stakeholders were selected using a combined purposive (as above) and snowball sampling approach (where stakeholders were asked to forward the survey invitation to relevant stakeholders). This survey yielded 107 responses, comprising 51 policing stakeholders and 56 delivery partners. Data was descriptively analysed.

2.4 Monitoring data

To provide additional context to the crime impact evaluation results, Tactical and Thematic Management of Risk in Law Enforcement (MoRiLE) scores and data on the number of community intelligence reports were collected. As this did not form part of the formal impact measurement, information on the data collected, analysis and results are in Annex D.

2.5 Limitations to the impact evaluation findings

The evaluation was designed to be as robust and comprehensive an approach as possible, within the parameters of the evaluation (such as budgets) and delivery progress of pilot sites. However, there are some key limitations which are important to bear in mind when considering the findings presented in the following chapters of this report. Key limitations relevant to both components of the impact evaluation are:

Delivery progress of CHB programmes: All sites included in the impact evaluation had been active for at least 6 months. However, as detailed in section 3.3, some sites were yet to implement all phases. As a long-term approach and given the entrenched issues in CHB areas, it is expected that CHB needs more time to have sustained impacts and it may be several years before impact is realised. This means that the results may have been somewhat diluted compared to an evaluation conducted in, for example, 2 to 3 years.

Contamination: CHB ran alongside other crime prevention initiatives which may have overlapped with pilot and control areas. Where specific location data for other interventions was available, this was considered in the analysis. For example, internal data found some pilot and control areas overlapped with Grip hotspot policing^{[footnote 5]}, and so an exercise was undertaken to map the overlap. This found that Grip was only active in a small part of the pilot or control areas, and so it is unlikely it had a large influence on results. The addition of a control variable to account for overlap between CHB and Grip was explored, although it was not implemented due to its inability to capture the nuances of Grip’s influence^{[footnote 6]}. It still remains possible that other initiatives reduced or increased the estimated impact of CHB on crime.

Estimating the average effect: It can be challenging to interpret average impact estimates, as it is often unclear what is driving the results. This is a general caveat of impact evaluation methods, which do not aim to understand the factors that may have influenced the success of a programme. In addition, CHB is a complex intervention with each programme unique in terms of phase delivery, its activities and crime context. Given the wide breadth of activities, multiple phases, and approaches to delivery, it is difficult to identify which elements of the programme had an impact on crime. However, the further analysis provides some indication of the contexts where CHB may be more or less effective.

Small sample: The evaluation included a small sample of CHB sites (based on the number of CHB sites in July 2024, however it is worth noting that at the time of the evaluation’s commission, the sample represented approximately two-thirds of CHB sites) and the further analysis split this sample into subgroups according to common characteristics. Due to the small number of sites in each subgroup, results are highly sensitive to the inclusion or exclusion of individual sites and may have limited the capacity of the evaluation to detect statistically significant results for the particularly small subgroups. Results from the further analysis should therefore be treated with caution and only used as an initial indication for where CHB may be more or less effective. In addition, crime displacement was assessed at site-level. It is therefore possible that some crime displacement was not detected due to low statistical power.

Interpreting increases in outcome measures: For some outcomes, an increase may be interpreted as a positive effect of CHB, particularly if they are observed during the Clear phase. For example, an increase in recorded drug-offences may reflect an increase in police activity. Alternatively, as outlined in the ToC (Annex G), a core outcome of CHB is an increase in the number of community intelligence reports. It is therefore possible that increases in crime may be a product of increased reporting or enforcement in CHB areas and, subsequently, detection of crime.

There were also some limitations specific to the community survey, which reduce the strength of conclusions that can be drawn from the survey results:

Unintended consequences: There is a risk that increased enforcement activity or media communication about police activity may have unintended negative consequences on pilot sites by drawing attention to crime and wider issues. This may have influenced the results, particularly the level of negative perceptions. For example, Koper et al. (2022) found that intensified levels of investigative and enforcement activity in an area were associated with more negative views of the police.

Representativeness: The community survey achieved a low response, introducing a high-level of non-response bias to the results. For example, low response rates introduce the risk that the survey respondents hold more extreme views than the sample that did not respond (Kibuacha, 2024). In this case, it may be that residents who are highly dissatisfied with crime and policing in their area were more likely to respond to the survey. In addition, there were limited means to increase the response rate (for example, through incentives or additional reminders) due to budget limitations. Although the data were weighted to better reflect the demographic profiles of each sample’s wider population, the results are unlikely to be representative and may be skewed towards more extreme views.

Evaluation timing: A retrospective approach to impact measurement was adopted because CHB had already been implemented in the selected pilot sites. However, this is an imperfect measure of change due to recall issues. For example, research shows that recent events (whether positive or negative) may influence whether someone thinks things have got better or worse over time because they cannot recall similar events that happened some time ago (Bradburn et al., 1987).

Influence of wider factors when measuring perceptions: Linked to the above, many of the outcome measures included in the survey are influenced by external factors that the evaluation cannot control for. For example, other HO evaluations highlight that perceptions of crime, safety and policing are also influenced by local and national media (Home Office, 2024a; Home Office, 2024b).

Nature of CHB sites: CHB is designed to focus on the highest harm and vulnerable locations which often have long-standing mistrust in the police and are areas of socio-economic disadvantage. Considering this context, the survey may have been conducted too prematurely for changes in perceptions to be realised. In addition, it remains a risk that there were some unobserved differences (such as levels of trust in police) between the pilot and control samples, which could be influencing results.

3. Overview of the CHB programmes

3.1 Interventions

Annex G shows the CHB programme-level theory of change (ToC), which sets out the intended outcomes and impacts of CHB, and the causal pathways for achieving these.

Analysis of the proforma identified at least 176 activities were being delivered through CHB, across the 12 evaluation sites that provided details on their interventions. The ToC demonstrates the broad range of activities being delivered, with 8 categories identified:

1. Strategic: activities to inform programme design or ensure effective delivery. This includes hotspot selection work, asset mapping, governance and multi-agency launch events.
2. Engagement: activities to inform and involve the local community in the CHB programme; examples include local surveys, media campaigns and events.
3. Enforcement: activities to monitor and disrupt OCGs; examples include high-intensity arrests, high-visibility policing, and the use of partners’ powers (such as Trading Standards to close illegitimate businesses).
4. Environmental: activities to improve the local environment; this includes fly-tipping removal, regenerating run-down spaces in the community and installing environmental blocks in crime hotspots.
5. Education: activities to improve knowledge of crime and exploitation; for example, educational programmes delivered in schools on exploitation.
6. Diversionary work: activities to prevent and deter people from engaging in crime; examples include the provision of leisure activities for young people, signposting to drug and alcohol services, or formal diversionary schemes.
7. Housing: activities to improve local housing or monitor tenancy compliance; for example, selective licensing schemes for privately rented properties and eviction notices for tenancy breaches.
8. Monitoring and evaluation: activities to monitor delivery status and progress against outcomes.

Section 5.1.4 provides examples of the types of interventions delivered in each phase.

3.2 Target SOC threats and crime types

In addition to the interventions, the proforma collected information on the SOC threats and crime types being targeted through the CHB programmes. All 13 evaluation sites ranked each SOC threat and crime type by whether they were high, medium or low priority, or not relevant. Full results are in Annex E; however, key findings are that:

• when asked about SOC threats, illegal drugs supply was most commonly ranked as high priority by pilot sites (11 sites)
• linked to this, when asked about crime types, drug trafficking and drug possession offences were most commonly ranked as high priority by pilot sites (10 and 9 sites respectively)
• cybercrime was ranked the lowest priority SOC threat, with 12 sites ranking this as low priority or not relevant
• ASB was a key issue targeted, with 8 sites ranking this as a high priority
• the following crime types were the lowest priority, with over three-quarters ranking them as low priority or not relevant: malicious communications, arson, residential burglary, commercial burglary, and miscellaneous crimes against society^{[footnote 7]}

3.3 Delivery status

The proforma also collected data on the delivery status of each site to understand their level of maturity. For the 11 sites that provided complete information, the delivery progress at the time of data collection was as follows:

• seven sites had completed or were delivering at least one iteration of each phase
• two sites had completed or were delivering at least one iteration of the Clear and Hold phases
• two sites were delivering Clear and were yet to start an iteration of Hold and Build

These results show that some sites were still in the early stages of delivery and had not yet operationalised the full framework.

4. Impact evaluation findings

This section presents the results of the crime impact analysis and community survey. Section 2.2 explains the methodology used, with additional detail in Annex A.

4.1 Crime impact analysis

4.1.1 Aggregate results

Table 2 presents the aggregated impact estimate for each crime grouping. It shows the estimated average effect size of CHB on crime rates (recorded crimes per 1,000 residents, per month) and the confidence intervals (CI) around this. The discussion focuses on the statistically significant results.

Table 2: Estimated impact of CHB on crime groupings (crimes per 1,000 residents, per month)

Crime grouping	Lower CI	Upper CI	Effect	p-value
Total PRC	-1.49	0.96	-0.26	0.67
ASB proxy	-0.34	0.37	0.01	0.94
Acquisitive crime	-0.49	-0.05	-0.27	<0.01*
Drugs	-0.20	0.26	0.03	0.81
Violence	-0.58	0.32	-0.13	0.57

Notes:

1. ‘*’ indicates statistical significance at the 95% confidence level.

At aggregate level, the evaluation detected a statistically significant reduction in acquisitive crimes. For a more tangible understanding of the impact of CHB, the effect size (-0.27) was scaled up using population data on the number of residents across all evaluation pilot sites. The analysis found that this result is equivalent to a 23.7% reduction in acquisitive crimes, or 51 crimes prevented per month in CHB areas, compared to the control group (CI: 4.6% reduction, or 10 crimes prevented per month to 42.8% reduction, or 92 crimes prevented per month)^{[footnote 8]}. In contrast, there was no statistically significant impact of CHB on any of the other crime groupings at aggregate level. In addition, the before and after analysis found no clear evidence of impact on knife-enabled offences.

The crime displacement analysis found no evidence to suggest that crime had been displaced to surrounding areas for sites that, at pilot level, saw a statistically significant reduction in crime.

4.1.2 Sensitivity analysis

Sensitivity analysis was conducted to test the stability of the aggregate-level results as the sample changed. Two types of sensitivity analysis were carried out:

• exclusion of single pilots: the analysis was repeated but removed a single pilot site from the sample each time
• reasonable best and worst scenarios: the analysis was repeated but removed the 2 best performing pilot sites (the worst scenario), and then again but with the 2 worst performing pilot sites removed (the best scenario)

The analysis found no instances where the exclusion of a single pilot (whether mapped at lower layer super output areas (LSOA) or output areas (OA) level) led to a difference in statistical inference. This means that, in all cases, there remained a statistically significant reduction in acquisitive crime and no statistically significant impact on any of the other crime groupings. This indicates that the results are not driven by a single site.

However, the reasonable best and worst scenario analysis suggests the aggregate-level results were driven by a small group of pilots. This sensitivity analysis found that, in the best case (2 worst performing sites removed), CHB resulted in a statistically significant reduction in all of the 5 crime groupings analysed. This finding can in large part be explained by the specific circumstances and implementation approach of the 2 worst performing sites that were removed from the sample. The process evaluation revealed that the first site excluded had not, due to resource constraints, been able to implement the full framework as intended, with limited governance and Build plans. The second site is inherently different to other CHB sites due to: its small resident but large transient population, which may make the Build phase more challenging to implement; its particular SOC threat focus; and a large increase in community intelligence reports seen after CHB began, which may have driven an increase in recorded crime. These factors may explain the more positive impact results when these sites are removed from the sample. In the worst case (2 best performing pilot sites removed), CHB had no statistically significant impact on any of the 5 crime groupings analysed.

These results demonstrate the potential of CHB in reducing crime across each of the crime groupings when implemented as intended and in the right contexts.

4.1.3 Further analysis

As outlined in section 2.2.1, further analysis explored whether crime impacts varied according to 3 types of site characteristics: CHB launch, implementation approach, and demographics. The key findings discussed below provide some insight into where CHB may be more or less effective, however other factors (such as implementation quality) may also be influencing the results. Annex B contains full results tables.

CHB launch

To understand whether specific rationales for introducing CHB in the selected hotspot were associated with better outcomes, the CHB launch analysis grouped sites by their motivation for implementing CHB. Interviews with police forces and data from the proforma indicated 3 main reasons for implementing CHB:

• as a reaction to a major crime incident in the area (such as a murder)
• OCG or harm mapping activity undertaken which identified the area as a location with SOC issues through data and intelligence
• deprivation or long-standing issues that have troubled the area

Results showed statistically significant reductions in all 5 crime groupings for the OCG or harm mapping group. No statistically significant impacts were found for the other 2 CHB launch groups. However, sites selected through OCG or harm mapping activity tended to be more mature than the other CHB launch groupings, running for an average of 22 months compared to 16 months for deprivation pilots and 12 months for major event pilots. To examine whether these findings were more likely linked to the selection approach or the maturity of site, a time control variable was included in the analysis. Broadly, the inferences drawn from the results remained the same. Pilots selected through OCG or harm mapping were still most effective, with statistically significant reductions in all crime groupings apart from drugs. On the other hand, no statistically significant impacts were found for the reaction to a major incident group, while statistically significant increases in ASB and drugs were identified for the deprivation group. Overall, these results suggest that CHB may be more effective in areas selected using a data-driven approach.

Implementation phase

To examine whether sites that were further along in their delivery progress saw greater impact, this analysis grouped sites by the phases implemented to date. Pilot sites were categorised by whether they had delivered:

• only a Clear phase so far
• a Clear phase and a Hold phase for at least 6 months
• a Clear phase, a Hold phase for at least 6 months and a Build phase for at least 6 months

The analysis found statistically significant reductions in acquisitive crime for the pilots categorised as conducting a Clear and Hold phase, or a Clear, Hold and Build phase, and the latter also had a statistically significant reduction in violence. No statistically significant impacts were detected for the Clear only group, however this group contained just 2 pilot sites. As outlined in section 2.5, this may have made detecting statistically significant impacts more challenging. Despite this, these results suggest that impact is more likely to occur if Hold and Build phases are carried out.

Area type

This analysis grouped sites by 2 types of area-based factors:

• area size (whether sites were small, medium, or large relative to the geographical pilot size of the sample)
• urban versus rural (whether sites were more or less densely populated)

The results showed CHB was most effective in the medium size sites (approximately 1.3 to 3.3km²), with statistically significant reductions in all 5 crime groupings analysed. On the other hand, statistically significant increases in violence were found for large size sites (approximately 3.3 to 15.7km²). As the ToC shows (Annex G), CHB also aims to increase community intelligence reports; it is therefore possible that this result reflects an increase in crime reporting in those areas instead of an actual increase in crime. This theory, however, could not be validated through monitoring data on the number of intelligence reports received due to inconsistent data from most sites. It may also be the case that resources are too diluted when implementing CHB across a wider area to have the desired impact. In terms of small areas (approximately 0.1 to 1.3 km²), there was no detectable effect. This may be partly due to the low monthly crime volumes in these areas, which made detecting statistically significant changes more challenging.

It is important to note that while the size classifications can be used to inform sites selection, as outlined in section 2.5, these were determined by a small sample of pilot areas, and therefore may not be generalisable to all CHB sites.

Statistically significant results were also detected for the urban versus rural analysis, with reductions in acquisitive crime, drugs, and violence for the urban site grouping. No statistically significant impacts were detected for the rural group. However, it is important to highlight that rural sites also tended to be larger and so this result might instead reflect the finding that CHB is less effective when covering a larger area.

Overall, as with the sensitivity analysis on best and worst scenarios, the further analysis suggests that CHB can be effective at reducing a range of crime groupings when implemented in the right contexts. Specifically, the evidence suggests that CHB may be more effective in medium size areas, urban areas, and when the site is selected through a data-driven approach. However, further analysis should be carried out to understand which of these factors, if any, predict crime reductions in CHB areas.

4.2 Community perceptions and resilience

4.2.1 Interpreting findings

As detailed under section 2.5, there are several limitations to the community survey results, which reduce the strength of findings and should be considered when interpreting the results. In particular: the low response rates across both the pilot and control sample (and subsequent risk that results are skewed towards those with more extreme views); the unintended consequences of enforcement activity; and sites being in the relatively early stages of delivery (especially the Build phase, activities during which are particularly important for achieving the outcomes measured through the survey).

4.2.2 Current perceptions

The community survey asked respondents to first consider how they felt about each factor on the day of the survey. Full aggregate-level results for the 5 pilot sites included in the analysis are in Annex C. However, compared to the control sample, key statistically significant findings on current perceptions are that pilot site respondents:

Have lower confidence and trust in the police: Both samples had poor levels of confidence in the police’s ability to prevent crime; however, this was worse among the pilot respondents. Nearly two-thirds (65%) of pilot respondents said the police were below average or very poor at preventing crime and disorder in their local area, compared to 50% of the control sample.

Have a weaker sense of community resilience and cohesion: Less than half (47%) of pilot respondents agreed people in their area get on well. This was lower than the control respondents (60% agree). The proportion of respondents who agreed that people in their area take care of each other or would help and/or intervene if they witnessed a crime was also lower among the pilot sample (respectively, 37% and 33% of pilot respondents agree versus 46% and 40% of the control respondents). It is important to highlight, however, that levels of agreement with the latter 2 measures were low, overall, in both samples.

Are less likely to view their local environment positively: Nearly one-third (31%) of pilot respondents agreed that their local area is a pleasant place to live. This was lower than the control sample (51% agree).

Have lower feelings of safety: The proportion of respondents who felt very or fairly safe walking alone in their area during the daytime was lower among pilot respondents (62%, versus 72% of control respondents). Both samples had low feelings of safety when walking alone during the night-time in their area. Again, this was lower among the pilot sample (23% felt very or fairly safe, compared to 37% of control respondents).

Have higher perceptions of crime: Perceptions of different crime types were fairly high across both groups. However, compared to the control sample, a higher proportion of pilot respondents said the following crime types were a very or fairly big problem in their local area: violence (44% of pilot respondents versus 36% of control respondents), burglary (54% versus 46%), drugs (82% versus 73%), fake goods (48% versus 34%), and weapons related crime (63% versus 47%).

These results demonstrate the challenging environments of the pilot site and control areas; however, across most metrics, perceptions were more negative among pilot site respondents than control respondents. Although the control sites had similar demographics to the pilot sites, key survey metrics (such as confidence and trust in police) could not be factored into the matching process due to data limitations. This should also be taken into consideration when interpreting the findings presented below in section 4.2.3.

4.2.3 Retrospective change

Following questions on current perceptions, respondents were asked how much each factor had changed compared to a year ago. Across the majority of metrics (nearly two-thirds; 19 out of 31) there were no statistically significant differences in the level of change reported by pilot respondents versus control respondents. However, for the remaining metrics, the proportion of pilot respondents who reported a negative change was significantly higher than control respondents in 9 metrics:

• confidence and trust in the police: effectiveness of the local police in preventing crime and disorder, and listening to the concerns of the public
• the local environment: the area being a pleasant place to live
• crime and safety: feelings of safety walking alone during the night-time, drug crimes as a problem and vehicle crime as a problem in the local area
• community resilience and cohesion: how well people take care of each other, trustworthiness, and likelihood to help and/or intervene if they witness a crime

In addition, the proportion of pilot respondents who reported a positive change was significantly lower than the control respondents in 6 metrics:

• crime and safety: feelings of safety walking alone during the night-time
• reporting crime: likelihood of reporting a crime experienced or witnessed to the police
• confidence and trust in the police: effectiveness of the local police in investigating crime, preventing crime and disorder, listening to the concerns of the public, and being visible in the community

It is important to highlight that in half (6 out of 12) of the metrics where significant differences were found, the majority of respondents in both the pilot and control samples reported no change compared to a year ago. In addition, it is possible that the higher proportions of negative change among pilot respondents may be a consequence of CHB drawing attention to crime and wider issues in the area through increased enforcement work and media reports on police activities, as explained under section 2.5. Similarly, it may be too early for any potential benefits of CHB to have flowed through to communities, especially from Build activities.

5. Process evaluation findings

This chapter summarises the findings from the process evaluation, focusing on key themes which arose from interviews with police and delivery partners from 6 pilot sites. It also reports key findings from the online stakeholder survey and additional insights from the proforma. See Annex F for full results from the stakeholder survey.

5.1 Implementation process

The HO stipulates several operational elements to effective implementation of the framework. This section focuses on the approach taken by sites to implement the key elements outlined in HO practitioner guidance on CHB.

5.1.1 Set-up and programme design stage

Hotspot selection

Identifying a hotspot for CHB is the first step set out in HO practitioner guidance, which recommends that intelligence and current knowledge should be used to select a high-harm area (particularly where OCGs live, operate and cause harm).

Most sites adopted a data-driven approach to selecting their CHB area, drawing upon both pre-existing knowledge of SOC hotspots in the force area and conducting a new analysis of data sources (such as data from wider agencies and PRC data). Participants felt it was important to draw upon a range of information sources to ensure a holistic understanding of the SOC threat and harm was generated and considered. A few sites identified their pilot site based only on pre-existing knowledge of high-harm areas, while a couple launched CHB following high-profile crime incidents (such as murders and firearm discharges) in the area linked to OCGs.

The selected areas were deemed suitable for CHB for a number of reasons. Generally, these sites had significant socio-economic disadvantages, which made them particularly vulnerable and tolerant to SOC. In some CHB sites, this manifested in the local community being reliant on counterfeit goods and services due to legitimate business being unaffordable to them. In others, there were large demographic groups (such as migrant workers or children in social care) within the community that were highly vulnerable to criminal exploitation by OCGs.

The infrastructure of the selected sites also increased their vulnerability to SOC. For example, cheap and poor-quality housing in these areas was often used to accommodate vulnerable people who were at risk of exploitation by OCGs (for example, through cuckooing^{[footnote 9]}). Alternatively, many uninhabited properties in these areas meant they were targets of organised acquisitive crime (such as burglaries for copper piping). Other types of infrastructure were also linked to increased SOC threat, such as transport links and the subsequent risk of county lines.

However, in some cases, the infrastructure of an area was a positive factor when considering its suitability for CHB. For example, some sites had existing partnership programmes active in the area which could be leveraged and merged with the CHB programme, reducing duplication. Other sites had a large availability of assets both within and surrounding the CHB area which could be used for the Build phase. For instance, large business chains which could be utilised for employment schemes, or football clubs to support with diversionary youth work.

Following hotspot selection, the practitioner guidance recommends that a locality profile of the selected area is produced to summarise its SOC threats. Most pilot sites created a locality profile as part of their CHB planning; however, its scope and purpose differed between sites. For example, a few sites created a locality profile of their wider force area to help identify the hotspot and reported finding it a useful tool. However, generally sites which created a locality profile did so of the pilot site location once it had been selected. Some sites did not produce a locality profile as part of the CHB set-up stage, mainly because an existing profile was available.

CHB team

Establishing a core team responsible for driving forward the implementation of CHB in the chosen area was an important step in the set-up process. Pilot sites identified 6 types of roles as key to programme implementation:

1. Senior responsible officer (SRO): The SRO adopted a strategic role in programme implementation and were generally of chief inspector or superintendent rank.
2. Lead responsible officer (LRO): The LRO adopted a project management role and was often a neighbourhood policing inspector in the CHB area.
3. Regional SOC community co-ordinator (SOC CC): The SOC CC adopted a strategic and facilitator role, sharing tactical guidance and acting as a conduit between the HO and forces.
4. Specialist in-force SOC roles: These adopted a tactical role, providing tactical advice and supporting with enforcement work. Examples included SOC disruption teams and local SOC CCs.
5. Analysts: These adopted a mapping and monitoring role of the CHB area.
6. Neighbourhood policing officers: Local policing teams adopted a supportive role, helping with intelligence gathering, on-the-ground delivery of enforcement work, and community engagement activities.

Although team membership varied between sites, participants largely held positive views of each role in their CHB team and valued their contribution and skills.

Community asset mapping

Following identification of the hotspot, the HO practitioner guidance recommends that sites undertake community asset mapping, which is an exercise where information is collated on the assets (such as key organisations, interventions and funding opportunities available) within the community that may be leveraged for CHB.

Sites used a range of sources to create their asset map. Most sites identified local knowledge as a key part of their asset mapping process, drawing upon the knowledge of individuals with strong links to the local community or area (such as the neighbourhood policing team). Asset mapping was also a collaborative process, with police forces utilising support from organisations they had existing relationships with. Although less common, some local organisations directly approached the CHB team to offer their support after becoming aware of the CHB project through word-of-mouth. In some cases, existing online repositories of local services were also utilised. Sites updated their community asset maps to ensure accurate information was retained.

Overall, sites viewed asset mapping as a useful process for understanding the existing resources which could be leveraged for CHB, in addition to identifying key gaps in local services. One site reflected finding the latter particularly helpful for monitoring the programme’s success, in terms of understanding whether their project filled key gaps.

Engaging delivery partners

The practitioner guidance provides limited information on engaging with delivery partners; however, a key recommendation was that a multi-agency induction or launch event should be delivered to key organisations identified during community asset mapping. In some cases, sites approached each organisation individually. This was most effective when trying to engage organisations which the police had not worked with previously. In these circumstances, face-to-face engagement was preferred to help establish rapport. In other cases, group engagement methods were used. These varied across sites; however, examples included small meetings, theory of change (ToC) workshops and, in line with practitioner guidance, large-scale engagement or partnership launch events. Group events were also used to maintain engagement. One approach involved holding a showcase event, which provided an opportunity to celebrate the achievements of the project and set goals for the subsequent year.

During the engagement process, most sites reported sharing information on the CHB framework and the SOC threat in the pilot area. It was felt explaining these elements in an accessible way was necessary to establishing a shared understanding of CHB, SOC and how delivery partners could support the project. This helped build consensus that CHB was a necessary programme for the identified area.

Across most sites, the local authority was a key delivery partner to the police, particularly staff from the housing, environmental, or youth services departments. Depending on the area, other key delivery partners were trading standards, housing associations, the NHS, local schools, and community organisations.

Selecting interventions

Consultation was an important part of intervention selection. All sites gathered the community’s perspectives, with a range of methods used, including surveys, workshops, meetings and participatory budget events (see section 5.2.2 for more information). In addition, most sites also described consulting their delivery partners. Generally, it was felt giving delivery partners the freedom to suggest ideas facilitated engagement and helped identify and fill gaps in services due to their local knowledge.

Several sites also drew upon data and intelligence. For example, local profiles and community intelligence helped identify the key drivers of crime in the area, therefore interventions were prioritised according to their relevance to those issues. Alternatively, tactical management plans (such as existing 4P SOC plans) were used to inform their disruption work and identify the most appropriate tactics.

5.1.2 Governance

The HO practitioner guidance sets out a three-tier partnership governance structure for sites to create. This comprises a strategic group, a tactical group, and several operational delivery groups (ODGs) organised around key themes (such as engagement, education, and environment). Most sites interviewed had a form of multi-tier governance structure in place which generally followed the HO recommended structure and was sometimes conceptualised as a ‘gold, silver and bronze’ structure, described below.

Strategic tier (gold)

Strategic tier meetings were attended by executive leaders within the police (such as the assistant chief constable) and delivery partner organisations (such as local authority chief executives); these met less regularly than other tiers, usually quarterly or as necessary.

The purpose of this tier was to monitor progress and ensure the CHB programme was achieving its overarching aims and objectives, in addition to planning for the future. A fundamental role was also allocating resources, given their authority, and to manage any significant barriers.

Tactical tier (silver)

Tactical tier meetings were described as an oversight tier for the operational level; membership of this tier varied between sites, although members needed to have sufficient decision-making power.

The purpose of this level was to organise and monitor the progress of on-the-ground delivery, and to review whether the programme was achieving the overarching objectives set by the strategic level. It was also responsible for helping the operational tier overcome any barriers they were experiencing in delivery, and escalating the issue upwards if it could not be resolved within the tactical level.

Operational tier (bronze)

Operational tier meetings generally consisted of multiple ODGs, thematically organised around the key issues in the area (such as the environment, housing, and crime).

Each ODG comprised frontline staff and were responsible for the on-the-ground delivery of the CHB activities, identifying any barriers to delivery, and escalating any issues that could not be resolved to the tactical tier. In some cases, the focus of each ODG was decided through consultation with the community and centred around the issues most important to the community while in others, these were decided in conjunction with delivery partners. Aside from ODGs with a crime or enforcement focus, these were mainly chaired by a non-police representative.

Sites adapted their governance structure as the programme progressed. For example, ODGs were merged which helped to embed a more problem-solving approach for ODGs discussing similar issues. One site adopted a flexible approach to ODG attendance by only inviting specific representatives if their attendance was relevant to the issues to be discussed, helping to minimise the burden.

Overall, sites valued the governance structure in place for providing accountability and embedding more structure in existing partnership work. In addition, the multi-tier model was seen to provide a clear escalation process.

5.1.3 Community engagement

As described above, sites used consultation methods to involve the community in programme design. Some sites created branding for their CHB programme and involved the community in the branding design. For example, some sites set up a competition for students in local schools.

To help maintain engagement, sites continued to consult the community throughout delivery. Consistent community meetings were held which provided regular opportunities for questions or concerns to be raised, or new issues to be prioritised. In addition, several sites emphasised the importance of police door-knocks, which provided further opportunities for the community to provide feedback.

Regular communication outputs were shared with residents. These provided updates on the programme’s progress (for instance, by reporting on successful disruption work) and advertised interventions. Sites used both physical (such as newsletters and leaflets) and digital (such as social media) channels of communication.

Sites were mindful, however, of how CHB was presented to local communities, publicising it as an approach to improve the local area. Sites avoided referencing the SOC focus of CHB to avoid creating fear among the community.

5.1.4 Delivery

Phase sequencing

Most sites started their delivery with Clear. In some cases, this was a natural starting point due to a high-profile SOC incident happening in the area which led to intensive operational work. In others, it was assumed that the framework should be implemented in the Clear, Hold and Build sequence.

As delivery progressed, phases were delivered concurrently. In particular, Clear often happened repeatedly and at the same time as Hold and Build. Sites reported that it is important to maintain a state of Clear throughout delivery and that delivering a single, short Clear phase was often seen as unrealistic. This was due to SOC being so entrenched within the CHB areas that significant time is needed to dismantle it.

Overall, sites adopted a flexible approach to phase sequencing. It was widely felt that phases do not, nor should, happen linearly due to every site being different and the police needing to adapt to new intelligence. However, a key reflection was that Build should begin early and before the Clear phase, to help maintain momentum throughout programme delivery and enable smooth transitions between phases.

Phase leadership

In all sites, Clear was led by the police due to the enforcement focus of this phase. However, delivery partners supported this phase by using their powers to help disrupt SOC activity from all angles. With regard to Hold, responsibility was shared between delivery partners and police in most sites. Build was mainly led by delivery partners due to their work and resources aligning with the purposes of this phase. However, police supported with some Build work to continue building trust with the community.

Clear

Clear interventions were mostly large-scale, intensive enforcement operations which involved a significant uplift in police resources into the CHB area. These interventions aimed to cause significant disruption to OCGs by arresting key nominals and seizing assets linked to criminality. Examples included:

• undercover operations focused on identifying properties, businesses and individuals linked to the exploitation of children and modern slavery
• days or weeks of action involving intense, high-visibility enforcement by police through the execution of warrants

As mentioned above, delivery partners also supported Clear by enforcing their powers, where appropriate, against people, groups and businesses that were linked to SOC or were having a negative impact on the CHB area. Examples included:

• Trading Standards inspections of business premises linked to illicit or counterfeit goods (such as illicit tobacco or vapes)
• Environmental Health inspections of business premises, particularly food outlets, suspected of being used for money laundering
• enforcement by environment officers against fly-tipping, litter, and graffiti

Hold

Hold interventions were largely preventative interventions that aimed to stop any further OCG presence or activity in the area after the initial Clear phase, and prevent wider issues from re-emerging after Clear. Interventions were fairly equally spread across police and partners. Examples of police-led interventions included:

• hotspot policing, ASB patrols and sporadic days of action to re-establish high-visibility and a strong enforcement presence in the area
• gang injunctions to prevent OCG nominals from returning to the area
• civil injunctions, such as Criminal Behaviour Orders (CBOs) or Community Protection Notices (CPNs)^{[footnote 10]}
• targeted advertising using technology to broadcast messages online encouraging individuals within a specific radius to submit intelligence

Examples of partner-led interventions included:

• situational crime prevention interventions, such as CCTV cameras
• environmental impact days, involving environmental health officers conducting house visits, fly-tipping checks, and clean-up days (including litter picking)
• enforcement work by housing officers, including issuing tenancy warning letters for breaches of tenancy or terminating tenancy agreements; this may be due to, for example, drugs being found in the property

Some interventions were focused on building public confidence and trust ahead of the Build phase. Sites described several community engagement activities, such as:

• creating a community hub or pop-up partnership surgeries, bringing together agencies (such as housing and alcohol and drugs services) in one place
• engagement stalls in busy public places within the CHB area (for example, outside a supermarket, collecting views on issues in the area)
• community meetings (such as coffee mornings) and workshops, which were sometimes used to set priorities for the CHB programme and inform the ODGs

Build

Build interventions were generally focused on tackling the drivers of crime to divert individuals involved in crime and prevent involvement in crime in the first instance, so that any reduction in SOC threat could be sustainable. Examples are:

• educational programmes delivered to parents on recognising the signs of child criminal exploitation
• educational programmes in schools – topics varied between sites but included confidence and motivation workshops, recognising the signs of exploitation and steps to take if they think they are a victim of exploitation, and reporting crime
• employment schemes with local businesses
• selective licensing schemes in the CHB area to tackle poor quality, privately rented housing; landlords who rent out properties in the area are required to obtain a licence from the local authority for each of their properties, which need to meet a set of conditions
• media campaigns on CrimeStoppers, including the youth equivalent ‘Fearless’, to increase awareness of the different ways to report crime
• diversionary schemes for young people identified as at risk of exploitation and engaging in criminality
• activities to increase the provision of youth activities in the area; examples included converting unused buildings into youth centres, creating local football teams, and providing free activities during school holidays
• mental health rehabilitation schemes

5.2 Factors affecting implementation

5.2.1 Partnership working

As a partnership approach, effective multi-agency working is a fundamental principle underpinning the successful implementation of CHB. The key elements of partnership working which affected the implementation of CHB are discussed below.

a) Collaboration and partner engagement

Early engagement

Collaboration between police and partners was a key facilitator to effective implementation. Police valued delivery partner support during the design stage for their local knowledge and often close connections to the CHB area. Involving delivery partners early on in the programme set-up (including in hotspot and intervention selection) facilitated long-term engagement in the programme.

Shared ownership

Establishing co-ownership of the programme early on was crucial and facilitated a whole-system approach during delivery. For Clear, this meant partners utilising their powers against OCG nominals to help disrupt the SOC threat. Although, partners first needed to understand how their capabilities could support with tackling SOC, and so educating partners early on was necessary. Some issues targeted through Clear were also not within police scope (such as fly-tipping), and so it was more appropriate for partners to take responsibility in solving these issues.

In terms of Build, police were involved in the delivery of some activities, which helped build trust with the community. Shared ownership was seen as fundamental to maintaining momentum in the long-term and achieving sustainability.

Low levels of partner engagement

Difficulties in securing engagement from some delivery partners created issues in implementation. For example, partners sometimes failed to dedicate the necessary resources to CHB. In one case, this resulted in the police leading delivery of the Build phase. Lack of partner engagement was mostly linked to the framework being seen as a Home Office or police initiative, or as a result of negative perceptions of the police. The latter was a particular issue when engaging community organisations, who were concerned that police support for the project would only be short term, or were apprehensive about being seen working with the police. This issue was overcome by leveraging support from mutual connections to help engage those organisations.

b) Awareness and understanding

Understanding CHB and SOC

Having a shared understanding of CHB helped to secure engagement and overcome a lack of buy-in. Emphasising the broader positive impacts CHB aims to achieve (such as improving the local environment) and sharing data on the impact SOC was having on the area facilitated this and helped to persuade partners of the framework’s value. However, the lack of understanding of CHB and its relevance to SOC among some delivery partners resulted in poor engagement. Future training or guidance on CHB for partners should look to improve their understanding of SOC, so that partners have a stronger understanding of how their powers can help to disrupt SOC.

Understanding the goals of the CHB programme

Positively, four-fifths (80%) of respondents to the online stakeholder survey strongly agreed or agreed that they and their collaboration partners have shared goals and expectations of CHB. Establishing shared goals ensured alignment between organisations and meant that all were focused on achieving the best outcome in line with the programme’s goals. This contrasts with previous ways of working, where organisations often viewed issues as outside of their responsibility. Formal agreements between police and partners which outlined the project aims and objectives created shared expectations for the programme. This helped to minimise conflict as the implementation progressed.

Understanding remits and capacities

Sites felt it was important to recognise that organisations will have other priorities and issues to manage alongside CHB, and their own processes. Awareness of those helped to manage expectations, facilitating the maintenance of good relationships.

Achieving widespread awareness

Widespread awareness of CHB within each organisation is needed to maintain momentum during the delivery stage. The lack of awareness of CHB internally (such as within key police force teams) and within other organisations (including the criminal justice system, such as the Crown Prosecution Service (CPS)) created delays as CHB-related work was not prioritised. In addition, it meant that additional internal resources which could have been leveraged were not. Ensuring an organisation had representation at each level of the governance structure facilitated a wider awareness of the concept. This provided scope for discussions on CHB to take place internally and separately to governance meetings, spreading awareness of the concept.

c) Communication

Effective communication

Four-fifths (80%) of respondents to the online stakeholder survey strongly agreed or agreed that they and their collaboration partners communicate effectively. This was fundamental to the delivery phase, especially with regards to intelligence sharing. For Clear, each organisation had access to different information that generated a stronger, more holistic intelligence picture when considered together. Sites emphasised to partners the importance of sharing information, even if the information seemed insignificant. Although, this needed to be a two-way process with the police communicating to partners the information sought. For Build, information sharing on vulnerable, at-risk individuals ensured they received appropriate support through referral to necessary agencies or diversionary schemes.

Effective cross-organisational planning

Good communication ensured effective planning of the delivery phase. Holding planning meetings with delivery partners ensured sufficient resourcing throughout Clear operational periods, with visible patrols from a range of enforcement agencies. For Build, good communication and planning between organisations meant that activities could be spread evenly over the week, minimising clashes. This maximised the accessibility of activities, increasing opportunities for community engagement.

5.2.2.Community engagement

a) Community involvement

Designing the programme around the community’s concerns

A second key principle of CHB is that decisions and solutions should be done with, not to, the community. Over four-fifths (85%) of respondents to the online stakeholder survey said that the community had, to some extent, been involved in the planning of CHB activities. Sites reported that community involvement facilitated engagement by ensuring their concerns, which were often considered low priority by police (such as fly-tipping or dog fouling), were understood. This meant that interventions which could address these issues were selected, encouraging engagement in the programme.

Using innovative consultation methods

Sites experienced challenges when consulting the community on programme design, sometimes struggling to reach their full target audience. For example, there was a lack of representation at community workshops or meetings from some groups and low response rates to surveys. Some sites used more innovative methods to overcome these common issues and empower the community to contribute to programme design. For example, a particularly successful method was a participatory budgeting event held in one site, which followed a ‘Dragon’s Den’ format where each applicant had 2 minutes to pitch their project to the community. The community then voted for their most favoured projects, and funding was allocated to projects with the most votes.

Acting on information and feedback from the community promptly

Sites felt that acting on community intelligence in a timely manner helped to build trust by demonstrating to the community that their reports would be taken seriously, encouraging further or additional community intelligence reports which were crucial for Clear and Hold. It also meant the community felt valued and listened to, helping to increase and maintain their engagement in the programme and its interventions.

Positive perceptions of community involvement

The community focus was seen as a key ‘value added’ of CHB, linking this to a reduction in SOC. For example, by tackling lower-harm issues that were important to the community, this was perceived to increase their trust and confidence in reporting more serious crime. Sites also valued the emphasis placed on improving community confidence in rejecting criminality. This was seen to help achieve a sustainable reduction in crime through communities becoming self-sufficient in maintaining a challenging environment for OCGs.

b) Reaching the community

Using a range of communication and engagement modes

Using a range of modes helped to maximise the reach of messaging. In particular, using a combination of physical and digital modes of communication meant that a broader range of groups could be reached. Although, some were less effective as they were not typically accessed by the community (for example, police webpages).

Face-to-face engagement with the community was also important. Community meetings were held at different times during the day to increase opportunities for engagement. Police visibility and attendance at activities (such as at youth clubs or community spaces) during the Build phase helped to develop more personal relationships with community members. This was linked to improved trust and, subsequently, increases in crime reporting. Their presence also provided opportunities for gathering intelligence from the community or through their own observations.

Delivering communication outputs in a timely manner

Some sites reported issues with delivering communication outputs in a timely manner, which hindered engagement. For example, delayed communications on participatory budget events meant some community groups did not have enough time to apply. Alternatively, restrictions on publishing results from enforcement activity before nominals are sentenced and CPS delays in reaching a charging decision meant success stories were delayed. This was seen to damage trust and may have led to a perception that the police are not taking action. The community may become apathetic to reporting intelligence if there are no visible and swift repercussions for nominals.

c) Negative perceptions of the police

Most sites identified mistrust as a key barrier to community engagement. This was seen to result from previous initiatives, which were often short term and left the community feeling abandoned once the project ended, and so the community was suspicious that this will happen with CHB too. This emphasises the importance of long-term commitment to the programme. In addition, some participants felt entrenched police mistrust meant the community was reluctant to champion a ‘police-led’ programme. To mitigate against engagement issues stemming from negative perceptions of the police, partners mostly led community engagement work while some sites purposely opted to exclude their police logo on any communications.

5.2.3 Resources

a) Financial resources

Filling gaps in services

There was a widespread perception that it is an unrealistic expectation for CHB to be delivered within existing budgets or, as HO practitioner guidance suggests, through alternative funding sources (such as other HO schemes, like the Safer Streets Fund or Violence Reduction Units). There were often gaps in existing interventions and services within CHB areas and while several sites sourced funding for new interventions through other schemes, this was not always possible. This meant that some programmes could not include activities relevant to key issues or community priorities. This challenge was supported by findings from the online stakeholder survey, which shows that nearly a quarter (23%) of delivery partner respondents strongly disagreed or somewhat disagreed that it was easy to successfully apply for funding to deliver CHB. This was incredibly disappointing for the local community and a source of frustration among practitioners.

Limitations of wider funding sources

Some of the key drivers of SOC in CHB areas (such as poor-quality housing) require substantial long-term investment and funding to improve. This made designing and delivering Build challenging, especially as identified funding sources were often short term which misaligned with the long-term ambitions of CHB sites. Similarly, some Build activities were delivered by third sector organisations, for which funding can be unpredictable and short term. This limited the sustainability of some activities.

Impact on staffing

The lack of bespoke funding meant that some activities were understaffed, resulting in changes to delivery plans. For example, Clear operations were deprioritised last-minute so that local officers could respond to other incidents. Had there been CHB funding, staff could have been ring-fenced to CHB through dedicated overtime, preventing delays. In addition, some sites strongly rejected the assertion that CHB is ‘cost-neutral’ due to the opportunity cost associated with diverting resources into the CHB area (and away from another area). Participants felt it is not sustainable to maintain concentrated effort in the CHB area, at the expense of other areas.

Sustainability benefits

The lack of bespoke funding was sometimes seen as a key facilitator to sustainability. Sites reported that this forces key agencies to undertake a comprehensive review of the existing resources in an area, helping to adopt a problem-solving approach by identifying gaps and maximising the capabilities of existing interventions. This ensures sites are working towards long-term achievements. Had bespoke funding been available, some participants felt this would have provided a negative incentive for involvement, and risk organisations from removing their support for the programme if funding was then withdrawn.

Considerations for future funding

There were several key considerations regarding any future bespoke CHB funding. For example, participants suggested this could be limited to interventions chosen by the community, or overtime for enforcement work only offered to officers involved in CHB or with the appropriate knowledge and skills.

b) Capacity

Dedicating enough time to CHB work

With the exception of a couple sites which had full-time CHB staff, most described CHB as an add-on to their normal duties. Participants reported this a significant strain on capacity and found it difficult to allocate sufficient time to CHB work, sometimes undermining the quality of outputs (such as the asset map). Alternatively, digital communication methods were prioritised for community engagement due to limited capacity for face-to-face engagement. This may have excluded some groups in the community (such as the elderly, or those without internet). More focused time for CHB would have helped maximise the quality of implementation.

Focusing on a small area

Many participants had wider-reaching responsibilities than the CHB area. For example, delivery partners often covered a whole local authority area, and SOC CCs covered several forces. This created challenges in implementation. For example, it was sometimes difficult to fully engage delivery partners as they did not have sufficient capacity to increase effort in the CHB area, while SOC CC support was spread too thinly. This may become more challenging as further sites are introduced.

Burden

Both police and delivery partners had to dedicate their personal time to the programme to complete CHB work. For example, meetings to engage delivery partners were often held outside of working hours to accommodate the operational hours of those organisations. There was also significant administrative burden associated with arranging, managing and co-ordinating actions from governance meetings. Overall, participants struggled to balance delivering CHB work with their normal duties.

Overcoming capacity issues through specialist staff

Some sites created dedicated CHB roles, or could draw upon specialist enforcement teams for support. These provided focused resourcing, which was seen to maximise the quality of implementation. For example, some sites assigned a specialist enforcement team to plan the disruption work and develop intelligence which helped the delivery of Clear and reduced burden on local neighbourhood officers. Alternatively, some sites had a full-time CHB lead, which provided a constant driver of the project, helping to maintain momentum.

However, it was sometimes felt that this approach to resourcing CHB could reduce sustainability. Sites may struggle to adapt to resourcing CHB within their normal duties if bespoke teams or staff are removed, which could result in the project breaking down.

c) Staffing issues

Staff turnover

Staff turnover disrupted momentum of the delivery phase in some sites, especially when positions were not filled quickly, and sometimes resulted in periods of inactivity. Staff turnover was seen as a key barrier to sustainability, particularly if committed individuals leave. However, handover periods between staff leaving and starting facilitated a smooth transition, minimising impact on implementation.

Small workforce

The small officer workforce of some police forces made resourcing CHB even more challenging, particularly for smaller or rural forces. These forces have comparatively fewer resources to draw from than larger metropolitan forces and so found it difficult to resource enforcement work. To overcome this, some sites took a phased approach to their Clear enforcement work, as opposed to an intense period of action.

d) Guidance

Gaps in guidance

Although the online stakeholder survey found that over four-fifths (84%) of respondents who said they received training ranked it as excellent or good, several gaps in the guidance resources available were identified by participants. Lack of information on the Hold phase seemed to be the biggest gap. Although participants understood the purpose of Hold, there was sometimes a lack of clarity on the activities that can be used, and how it is distinct from the other phases.

Other gaps mentioned were guidance on identifying a hotspot, accessing funding, ‘exiting’ CHB, monitoring and evaluation, asset mapping, and developing effective communications for the community. For hotspot selection, some participants felt that an analytical guidance document explaining how to go about identifying a hotspot, including tips on the use of data, the MoRiLE tool, information sharing agreements (ISAs) and good practice examples of how other pilots were picked, would help forces with fewer or less experienced analysts. Overall, more structured, formal guidance or learning from the earlier pilot sites on the stages to implementation would have been valued. These points provide important learning for future guidance.

The value of guidance and support

Despite the identified gaps in guidance, all sites still made positive reference to the resources or support provided. Participants noted a range of guidance they found useful, specifically training sessions on CHB delivered by the HO, presentations, LRO guidance, and learning from other sites. These helped stakeholders understand the aims of CHB and provided examples from other sites for inspiration.

5.2.4 Knowledge and influence

a) Knowledge and skills

Those involved in CHB needed to have the appropriate knowledge and skills to facilitate effective implementation. Local knowledge was fundamental to the LRO role, as they had knowledge of key delivery partners within the community and the OCGs. The latter was seen as crucial to the planning of disruption work. However, some participants felt that they may not have the sufficient knowledge to manage the OCG threat or understand the link between some of the activities and reduced SOC, due to their lack of specialist SOC training. Improving guidance resources was seen as a key solution to overcoming these gaps in knowledge.

In contrast, knowledge of SOC was a key strength of the SRO, SOC CCs and specialist in-force SOC roles. Their knowledge of SOC disruption tactics helped to inform tactical planning. SOC CCs in particular were valued for their extensive knowledge of SOC, especially as they had access to a range of disruption tactics used across their region and nationally. They facilitated learning across sites, sharing best practice examples, which provided confidence to sites in the effectiveness of their tactics. Overall, SOC CCs were described as an “integral” part of CHB.

b) Power and influence

Those involved in CHB also need to have an appropriate level of power and influence, particularly those involved in governance. Some sites had their tactical tier members as chairs of the ODGs, as these were often managers or more senior members of staff. It was felt that senior presence provided more accountability to ODG representatives by ensuring they attended and provided updates as expected. However, for governance to be effective across all levels, representatives need to have sufficient decision-making power or capability to deliver and help solve issues.

In some cases, the lack of influence or power of those involved was a barrier to implementation. For example, CHB leads generally did not hold sufficient rank to sign-off on key decisions, sometimes causing delays. Similarly, some SROs reported that they did not have the authority to divert resources to CHB. In addition, SOC CCs found it difficult to secure senior officer time during force engagement, which they felt stemmed from their lower rank.

5.2.5 Senior support

a) Resourcing

Senior support was crucial to ensuring the programme received adequate resource. At the force executive level, they had the necessary influence to prioritise CHB and divert resources from across the force into the CHB area. Senior support also meant those involved in the day-to-day running or on-the-ground delivery were given sufficient time to focus on CHB work, without being pulled away for other demands.

b) Engagement

Senior support during the delivery partner engagement process helped increase the perceived importance of the project, thus encouraging buy-in from delivery partners. However, it was felt that senior support needs to be obtained early so that they can drive buy-in within their organisation. Senior support also helped maintain momentum in implementation, as their support could be leveraged if there was any pushback from frontline staff. Overall, senior support was seen by some as key to the sustainability of CHB and long-term engagement in the programme.

c) Lack of practical support

Overall there appeared to be good senior support for CHB, as demonstrated by the online survey results which found that nearly 9 in 10 delivery partner (88%) and police (86%) respondents strongly agreed or agreed that they had the right level of senior leadership buy-in and support. However, some felt that seniors provided inadequate practical support for the programme by, for example, failing to establish internal agreements with relevant teams (such as forensics) to prioritise CHB work.

5.2.6 Existing relationships, resources and structures

a) Minimising burden and duplication

Leveraging existing resources and structures helped to reduce duplication and minimise the burden on those involved. For example, using existing repositories of local services made the asset mapping process more efficient. Alternatively, one site leveraged existing partnership meetings for their governance. This involved creating a single ODG consisting of key representatives from existing problem-solving groups for the area, as opposed to a set of thematic ODGs. This helped to minimise duplication while still providing a mechanism for CHB to feed into partnership meetings.

b) Facilitating contact through mutual connections

Leveraging existing relationships was crucial for identifying and engaging delivery partners, particularly community-led organisations. Sites used support from delivery partners with strong links to the CHB area (for example, as residents of the area) to facilitate contact with these organisations, who may hold negative perceptions of the police. Referral through a mutual and trusted connection provided reassurance.

5.2.7 Intelligence and data

a) Being intelligence- and data-driven

Ensuring an appropriate location for CHB

Sites reported that adopting a data-driven approach to hotspot selection ensured an appropriate location was chosen due to potential locations being assessed consistently and objectively, and validated areas provisionally selected through less formal evidence. These findings align with the impact evaluation results (section 4.1.3), which suggested CHB was more effective in sites selected through data.

Data or intelligence-led interventions

Using data and intelligence collated in problem profiles and existing 4P SOC plans enabled the most appropriate interventions to be selected to address the key issues and SOC threats in the area. Reviewing intelligence throughout delivery helped sites to deliver new interventions which could target issues as they arose.

Generating intelligence through a range of methods

Strong intelligence was a key facilitator to the delivery of Clear and Hold. Sites reported that generating intelligence from a range of sources (such as CrimeStoppers, covert operations, and from partners) was fundamental to effective disruption during Clear and meant the police could be confident that arrested nominals would be charged. During the Hold phase, continual intelligence monitoring meant resources were focused on the most appropriate locations and individuals of interest.

b) Data limitations

Boundaries

The boundaries of potential CHB hotspots often did not align to the geographical boundaries of data. This was particularly challenging for sites which planned to use partner and police data simultaneously, as boundaries between sources often did not align. In addition, while it was easy to identify the location in which an OCG’s activity was concentrated, their crime activities impacted wider areas. These issues made identifying a clear boundary for some sites challenging.

Data gaps

Some sites reflected that access to additional partner data, such as on school attendance (education) or on counterfeit goods (trading standards) would have been useful during hotspot selection. This would have provided a more holistic view of the CHB area and its vulnerabilities and provide stronger evidence of key issues.

5.2.8 Practical considerations

a) Bureaucracy

Data sharing

The processes associated with data sharing documentation created challenges in implementation. For example, some sites found setting up data protection impact assessments (DPIAs) or ISAs a lengthy process. Alternatively, some sites failed to have these documents in place at all which limited information sharing. These issues hindered the hotspot selection process due to delays in accessing partner data, and damaged inter-organisational relationships during delivery. Different IT systems and additional security measures of police systems limited access to datasets, which made sharing data challenging and was described as a key source of inefficiency.

Maximising accessibility of funding applications

Sites highlighted that funding application processes often exclude community-based organisations as a result of, for example, language barriers or insufficient paperwork. Participatory budgeting helped remove these common barriers to accessing funding by simplifying the application process. While participatory budgeting was viewed positively, finance processes delayed the allocation of funding to successful local groups, damaging the relationships between police and those groups.

b) Size

Area size created challenges during the community asset mapping process. For example, a particularly small CHB area found there were not sufficient assets, which hindered their development of a comprehensive asset map and, subsequently, the site’s ability to design an effective Build programme. This may also explain the lack of statistically significant impacts found for sites categorised as ‘small’ (see section 4.1.3), as possibly these areas were not suitable for CHB to begin with.

5.2.9 Perceived effects of CHB

a) Improved partnership working

More and better collaboration between agencies

Participants felt CHB has created a shift from siloed to collaborative working practices, strengthened existing relationships and facilitated the development of new ones, and integrated the police into existing partnership work. This positive sentiment was supported through the stakeholder survey, which found that nearly all respondents (93%) strongly agreed or agreed that CHB has fostered collaboration between the police and delivery partners in tackling SOC and wider issues in the community.

Helping to create a whole-system approach to tackling SOC

Overall, sites valued CHB for the framework it provides in integrating the police, partners and the community in tackling SOC, and helped forces focus on the Protect, Prepare and Prevent elements of the 4P framework. CHB was seen to provide a “wrap-around” approach that means all agencies are involved in improving the area and removing the impact of SOC, as opposed to a police-focused approach which tends to temporarily remove immediate SOC threats.

b) Progress against outcomes and impact

Clear

There was strong consensus that the delivery of Clear was successful, and sites described achieving a number of Clear outcomes. For example, all sites reported a reduction in SOC or OCG activity, and most described the removal of OCG nominals. One site reported that, as a result of CHB activities, there was no active OCG in the area or open county line^{[footnote 11]}. Most sites also seized assets and illicit goods, sometimes seizing substantially more than anticipated.

Hold

Anecdotal evidence suggested that some progress had been made against Hold outcomes. A few sites described that the community now feel listened to and some also felt that there had been improved engagement, evidencing this with examples of increased involvement in community activities (such as litter picking). In addition, a common outcome reported was that the area had become safer. Participants linked this to an increase in visible police presence in the area providing reassurance, and introducing CCTV. While sites described their ongoing activities to reduce the risk that OCGs fill the void created through Clear, this was not explicitly referenced as being achieved.

Build

Some sites reported that Build had gone reasonably well, however others reported struggling with this phase or that they were too early into their delivery to comment on the success of this phase. Sites described making progress against a few Build outcomes. Most sites reported that the community has greater pride in and responsibility for their area and that there was an increase in the number of community intelligence reports received. Participants felt this was due to visible police action, which had subsequently improved public confidence and trust in local policing and therefore increased the community’s willingness to report crime. Although improvements in these outcomes is not evidenced by the community survey results presented in section 4.2.3, participants provided anecdotal evidence of increased trust in the police, illustrating this with examples of gratitude from the community such as thank-you letters and posts on social media.

In relation to the programme’s overall impact, most sites perceived that crime and SOC-related activity had reduced in the area. For example, results from the online stakeholder survey showed that around 7 in 10 respondents (71%) strongly agreed or agreed that CHB has reduced the SOC threat in the area, and nearly a quarter (74%) strongly agreed or agreed that it had reduced general crime in the area. Evidence provided during interviews to support this perception varied. For example, some reported reductions in recorded crime, while others provided anecdotal evidence of writing fewer community impact assessments^{[footnote 12]}. Most sites also reported a reduction in their MoRiLE scores since the inception of CHB, directly linking this to CHB activity. Some sites reported a reduction in ASB incidents, illustrated through community reports, recorded ASB, and anecdotal evidence.

6. Conclusion

6.1 Summary of impact evaluation findings

Overall, this evaluation has shown that CHB can be an effective approach for reducing crime when implemented properly and in the right contexts.

At aggregate level, the evaluation shows a statistically significant reduction in the acquisitive crime grouping (comprising robbery, residential and commercial burglary, and theft of a vehicle). The effect size was equivalent to a 23.7% reduction, or 51 acquisitive crimes prevented per month across CHB areas, compared to the control group^{[footnote 13]}. However, there was no statistically significant impact on total PRC, the ASB proxy, drugs, or violence. These results represent the average impact of CHB.

As demonstrated through further analysis, CHB was more effective when implemented in specific ways or settings. Key findings were that:

• medium-sized sites (approximately 1.3 to 3.3km²) were most effective at reducing crime, with statistically significant reductions found in all 5 crime groupings analysed
• although likely related to the site size, urban sites (areas with greater population density) were more effective at reducing crime than rural pilots; statistically significant reductions were found in 3 crime groupings analysed (acquisitive crime, drugs, and violence)
• sites which adopted a data-driven approach to hotspot selection through OCG and harm mapping were most effective at reducing crime, with statistically significant reductions in all 5 crime groupings analysed

While these findings should be interpreted with some caution due to the small sample of sites included in the evaluation and the complexity of identifying which factors are driving results, the process evaluation research provided additional context which may support these findings. Resourcing issues were a key challenge experienced by sites, with significant capacity and staffing issues widely reported. Therefore, it may be that medium-sized sites achieved a sufficient level of intensified resources, without spreading resources too thinly. On the other hand, it may be more difficult to detect statistically significant impacts in small sites as a result of statistical power issues. In addition, small pilot sites were reported to have limited assets to leverage, which made developing a comprehensive Build phase challenging. Regarding site selection, there was widespread consensus that adopting a data-driven approach is effective, providing an objective assessment of an area’s appropriateness for CHB.

Evidence of impact on community perceptions and resilience is less positive. Overall, the results emphasised the challenging areas within which CHB is operating, with particularly low levels of confidence and trust in the police, a weak sense of community resilience and cohesion, and relatively negative views of the local environment. As highlighted through the process evaluation, these perceptions are complex and may take a long time to change.

Across most metrics, the survey found no statistically significant differences in the level of change reported by respondents in the pilot versus control areas. However, in some cases there was evidence of negative impacts, with respondents in pilot areas significantly more likely to say that things had got worse. These negative changes covered metrics under community resilience and cohesion, the local environment, and perceptions of crime and policing. However, it is important to consider these results in relation to key limitations which may be influencing these. The most important limitations are:

• the survey had a low response rate (less than 2%): this increases the risk of non-response bias, and it may be the case that the survey respondents hold more extreme views than the sample that did not respond; this reduces confidence in the representativeness of the survey, with a risk that the results have been skewed towards more extreme views
• there may be unintended negative consequences of CHB as a result of increased enforcement activity or media communication about police activity that may be drawing attention to crime and wider issues; this may have influenced the results, particularly the level of negative perceptions
• although control sites had similar demographics to pilot sites, the matching process could not take into account key survey metrics due to data limitations (such as trust in police)
• due to evaluation timing, a retrospective approach to measuring change was adopted; however, this is subject to recall issues, limiting the accuracy of findings
• most outcomes measured by the survey are expected to be achieved through Build activities: however, several sites were in the relatively early stages of Build when the survey was conducted, or were yet to begin their Build delivery; thus, the survey may have been conducted too prematurely to see any positive impacts

6.2 Summary of key process evaluation findings

Overall, sites valued CHB for the framework it provides in embedding a multi-agency and 4P approach to reducing SOC and its harms. There was strong consensus that CHB has improved partnership working, creating a shift from siloed to collaborative working practices. However, perceptions on the success of delivery varied by phase. Sites described making good progress in outcomes for Clear, but only some outcomes for Hold and Build. There was a strong understanding of the framework among both police and delivery partners.

HO practitioner guidance sets out several operational elements to effective implementation of CHB. Generally, sites seemed to follow these recommendations. For example, most sites adopted a data-driven approach to selecting their CHB area and had a multi-tier governance structure in place.

Sites adopted a collaborative approach to implementation. During the design stage, police forces utilised support from key partner agencies to collate information on the key organisations and interventions in the CHB area (called an asset map) and identify gaps in local services. Sites also consulted the community to help select interventions to be delivered through their CHB programme, facilitating engagement. During the delivery stage, police and delivery partners were involved in each phase, though Clear was mostly led by the police and Build by partners.

Phase sequencing was flexible, with phases often running concurrently. Although most sites started their delivery with the Clear phase, it was felt that delivery and phase sequencing should be flexible, driven by intelligence. In hindsight, some sites felt starting their delivery with Build activity may have been a more effective approach, enabling seamless transition between phases.

The process evaluation identified a set of key themes, which covered cross-cutting facilitators and barriers to implementation. Key facilitators to implementation included adopting an intelligence and data-driven approach, particularly to programme design.

In addition, leveraging existing resources, relationships and structures was seen as valuable for minimising duplication, maximising partner engagement, and reducing burden. Furthermore, senior support was reported as essential to engaging delivery partners and helping to maintain momentum. However, senior support seemed most important to ensuring the programme was adequately resourced, particularly when the core CHB team did not have the power to make decisions on resource allocation.

There were 2 fundamental principles of CHB which underpinned its successful implementation, specifically effective partnership working and community involvement. With regards to effective partnership working, this was facilitated through a number of mechanisms, including: co-ownership of the programme; shared understanding of CHB, SOC and each organisation’s capabilities; a multi-tier governance structure to facilitate escalation of issues, communication and accountability; and good communication through information sharing between organisations during delivery.

In relation to community involvement, consulting the community early in the design stage was key to ensuring the programme could be designed to address their concerns and subsequently maximising their engagement in the programme. However, community engagement during the delivery stage through a range of mechanisms was also seen as key to maintaining their engagement and facilitating community intelligence reporting.

Although there were several key barriers identified, there were no critical challenges which caused a significant derailment in implementation. Key barriers included resourcing issues, particularly capacity constraints, lack of bespoke funding, and gaps in CHB guidance, which sometimes caused delays in delivery and undermined the quality of implementation. Low community engagement and, in some cases, lack of buy-in from delivery partners created challenges in the delivery of the Build phase and disrupted momentum. In addition, there were some bureaucratic issues which created challenges in delivery, such as different IT systems between organisations which made information sharing more difficult, or unanticipated financial processes which delayed the allocation of funding. Working to overcome barriers as they arose was key to maintaining momentum and progress in delivery.

6.3 Overall learnings and recommendations

6.3.1 Implementing CHB successfully

The following key learnings for future implementation of CHB have been drawn from across the impact and process evaluations, to maximise its success:

• formal agreements should be established between organisations involved in implementation, which can help collaboration, facilitating the creation of shared goals and expectations, and the process of information sharing
• using data to inform programme design is key to ensuring the most appropriate hotspot is selected; the activities chosen can address the target issues, and there is shared awareness of the key problems in the community
• co-ownership between police and partners should be established early on, so that agencies support each other throughout delivery and are part of each phase; involving partners in the programme design and educating partners on their relevance to SOC is key to achieving this
• sites should be considerate of the level of resource required to implement CHB at the outset and so should ensure enough resource is dedicated to the programme; where possible, senior staff should ringfence resource to CHB
• linked to the above, senior staff should give practical support to the programme where appropriate; for example, by securing internal agreements for relevant teams to prioritise CHB work
• widespread awareness of CHB within each organisation helps to acquire additional resources for the programme
• a multi-tier governance structure should be established to help keep the project on track by providing accountability and a clear escalation structure for issues to be resolved swiftly
• where appropriate existing relationships, resources and structures should be leveraged to minimise duplication and maximise efficiency
• good information sharing is vital for the delivery of interventions in all phases; using technical solutions such as MS Teams provides an efficient mode for sharing information between organisations
• a range of community engagement methods should be used to maximise reach, particularly to hard-to-reach groups
• the community should be consulted early in the design stage to ensure their views are captured and acted upon, which is key to achieving and sustaining engagement
• improved guidance from the HO on CHB would be beneficial, particularly with regards to identifying hotspots, asset mapping, and the Hold phase
• sites should be mindful of internal processes which can cause delays and disrupt momentum – effective planning could mitigate challenges associated with this; for example, a clear financial process should be communicated to applicants of funding to manage expectations
• the size of a hotspot is important and should be considered – an area needs to have sufficient local amenities and assets to leverage to support the development of a comprehensive Build phase, but an area should not be too large, as this dilutes resources
• sites should adopt a flexible approach to delivery; phases can run concurrently and there is value in starting with Build activities first
• the government should consider providing bespoke funding for CHB; the eligibility criteria would need to be carefully considered to overcome common challenges associated with other funding streams (such as the short-term nature of funding) and prevent undermining the sustainability of CHB (which is a risk if funding were to be removed)

6.3.2 Future evaluation

The following recommendations for future evaluations have been identified:

• regression analysis should be carried out to understand which factors predict crime reductions in CHB areas (for instance, pilot site, population density, and level of site maturity)
• a community survey should only be conducted where budget and timings allow for more robust data collection, such as in future CHB sites before police activity begins and then repeated at appropriate intervals to properly measure changes in perceptions; in addition, alternative surveying modes (such as face-to-face surveying conducted by independent fieldworkers) should be explored to help achieve a better response rate
• if a robust community survey is not feasible, alternative evaluation methods for understanding the impact of CHB on community perceptions and resilience should be considered; for example, other HO evaluations used a form of contribution analysis called the Qualitative Impact Protocol, which explored changes in perceptions of safety through in-depth interviews (Home Office 2024a; Home Office 2024b)
• an economic evaluation through a robust cost-benefit analysis should be undertaken, to understand whether CHB is value for money
• it could be beneficial to carry out follow-up analyses to identify any future impacts of CHB on crime, when more sites have implemented the full framework and had more time to embed the Build phase

Annex A: Technical annex

This annex provides a more detailed methodology for the impact evaluation.

Technical methodology: crime impacts

The impact evaluation was designed to maximise analytical robustness as measured by the Maryland scientific methods scale (MSMS). The evaluation of CHB’s effect adopted a quasi-experimental approach, comprising 4 stages, each discussed below. The methodology enabled the analysis to be conducted at an MSMS level 3 by providing a statistically matched, non-treatment comparison group.

Stage 1: Propensity score matching (PSM)

Crime rates are influenced by a complex interplay of factors, including socio-economic conditions. The association between socio-economic conditions and 12 different types of crime were analysed to understand which factors best predict high crime rates. These are the factors that were taken forward for the PSM, so that pilot areas could be matched to similar areas that were not affected by CHB.

Data on socio-economic conditions was sourced from the 2021 Census. The socio-economic conditions are described by 7 variables, specifically: the number of white English residents, numbers of people living in high deprivation, the number of residents in a given area, country of birth (whether a resident was born in the UK), the numbers of people that are economically active, the number of residents in a higher management position, and the number of residents without qualifications. Gender was also trialled, although this did not have a statistically significant impact on crime.

An ordinary least squared regression was estimated for each type of crime. The analysis was conducted in advance of data collection from the HODH and, as a result, used police.uk datasets covering every force for June 2023. The types of crime that were predicted from the regression broadly match with the crime types to be analysed through the impact analysis, and were as follows:

• antisocial behaviour
• drugs
• bicycle theft
• burglary
• criminal damage and arson
• other crime (includes forgery, perjury and other miscellaneous crime)
• possession of weapons
• public order
• robbery
• theft from the person
• vehicle crime
• violence and sexual offences

For total crime, each of the 7 socio-economic variables described above were statistically significant at the 1% level, indicating these are good predictors of crime. These factors are backed up by the academic literature. For example, Vilalta-Perdomo et al. (2021) found communities with high crime rates exhibit the following: economic deprivation, inequality, population size, density and overcrowding, high population mobility and high non-white population.

Once these variables had been selected, PSM was employed to match pilot sites with non-treatment areas within the same police force. The PSM methodology was developed by Rosenbaum and Rubin (1983) and attempts to balance the distribution of measured baseline covariates between treated and untreated subjects, such that the only difference between the 2 groups is exposure to some kind of treatment. In this case, the treatment is introducing CHB. The conditional-independence assumption requires that the common covariates that affect treatment assignment (whether or not the site is selected for CHB) and treatment-specific outcomes (crime rates) be observable and that there are no unobserved factors which affect treatment.

The propensity score is estimated by a logistic regression model, in which a treatment indicator is regressed on observed baseline characteristics. The estimated propensity score is the predicted probability of treatment derived from a fitted regression. A propensity score model was applied in each police force area to find suitable control areas to be compared to the pilot areas. To eliminate the tendency to find control areas that lay geographically adjacent to the pilot area (due to similar demographics), an exclusion zone was defined around the pilot. All LSOAs within this zone were eliminated from the matching pool.

Table 3 below shows the distribution of covariates before and after matching for all areas to show the success of the propensity score approach in balancing the distribution of covariates shown to influence crime rates.

Table 3: Balance of covariates before and after PSM

	Std. mean difference before PSM	Std. mean difference after PSM
White English	0.9	-0.2
Usual residents	1.6	-0.1
Deprivation	-0.2	0.1
Country of birth	1.5	-0.2
Economic activity	1.4	-0.1
Higher management	0.4	-0.2
No qualifications	1.6	0

Stage 2: Synthetic controls

Given that the PSM process did not account for crime rates, there was an opportunity to further enhance the match between pilot and counterfactual through the synthetic control method (SCM). The SCM aims to align pre-intervention crime trends between the pilot and counterfactual sites by weighting the control group, thus ensuring a closer match. The SCM was first introduced by Abadie and Gardeazabal (2003) and then further developed in Abadie et al. (2010) and provides a method to analyse the impact of place-based interventions such as CHB. These studies typically have a single unit (often represented by a city or region) that experiences, at a point in time, an event (or treatment) while the rest do not. To evaluate whether the treatment had affected some outcomes in the treated unit, relative to what would have happened in its absence (the control), the method identifies a control called the synthetic control unit.

In this evaluation, LSOAs were indexed by J = 1, 2, 3… S+1, where pilot areas were denoted as 1, and the S potential counterfactuals – the donor pool – were denoted from 2 to S+1. The SCM then identified the vector of weights W* = (w₁,w₂…w_S) to approximate as closely as possible the pre-treatment crime rates in pilot LSOAs. Once W* had been identified, it was used to calculate the post-treatment outcome variables for the synthetic control units by weighting each LSOA and OA appropriately. The SCM added value by improving the match between pilot and counterfactual group.

Stage 3: Parallel trends

Following implementing synthetic controls, 2 tools were employed to assess the quality of the match between the pilot and the synthetic control:

• a visual inspection of pre-intervention trends
• an event study

The visual inspection provided an initial indication of the success of the SCM. Visually, 6 sites appeared to have achieved a good SCM match, 5 an average SCM match, and 2 a poor SCM match. The visual inspection considered both crime levels and crime trends and focused on the long-term trajectory of crime in the pilot and counterfactual areas. To reduce the period-to-period variation in the data, crime rates were visualised on a 3-month rolling average.

To improve confidence in the conclusions drawn from the visual inspection, an event study was conducted for each pilot site. An event study demonstrates whether there is a statistically significant intervention effect before an intervention is applied. If the parallel trends assumption is satisfied, the intervention effect at any point before the intervention is applied should not be statistically different from zero. If there is a statistically significant impact before the intervention, then this reduces the extent to which any post-intervention effects are attributable to the intervention. If an intervention has a significant impact, then time periods before the intervention should show no effect, and time periods after should show an effect. The event study was completed via the regression specification detailed below:

Time_t is a set of T dummy variables, one for each time period of the data: Time_t is equal to 1 if t is equal to the time period the dummy represents. Pilot_i is a binary variable to refer to either the counterfactual (0) or pilot (1) group, for observation i. The coefficients of interest are β₃, _t for t = 1 to t = T: these represent the effect of the pilot within a single time period. Specifically, for the time periods before CHB began, if the coefficients are statistically significant from zero, then this indicates that parallel trends are not satisfied. The event study identified 3 sites with pre-intervention time points that were statistically significant from zero. This reduced confidence in the control match. To reduce any residual pre-intervention effect as a result of differences in crime trends, the analysis was conducted with a DiD approach, detailed below. Figures 1 to 5 present the pre-intervention crime trends for the pilot sites and synthetic control group, for each crime grouping.

Figure 1: Crime rates for total PRC before CHB

Figure 2: Crime rates for ASB proxy grouping before CHB

Figure 3: Crime rates for acquisitive crime grouping before CHB

Figure 4: Crime rates for drugs grouping before CHB

Figure 5: Crime rates for violence grouping before CHB

As the SCM was applied to total PRC, there is a closer match between the 2 groups in this metric than in the others. In particular, there was some divergence in the levels of acquisitive crime. Crime levels in the control group were declining before implementing CHB, while crime levels were increasing in the pilot group. As the pre-intervention trends are less similar than desired, the assumption that the control group indicates what would have happened in the pilot group in the absence of CHB is weakened. It is possible the results have been biased towards an intervention effect due to deviations in pre-intervention trends, which means that results for acquisitive crime should be interpreted with some caution.

Stage 4: DiD approach

The analysis used a standard DiD linear regression specification, as follows:

In this specification, the DiD variable is the interaction between Pilot_i – a dummy for whether the LSOA is in the pilot or control group – and CHB CHB started_i; which is a dummy for whether CHB has been rolled out in the relevant pilot at the time of observation. The DiD specification did not use clustered standard errors, which are often used when there is heteroskedasticity. Consequently, there is a caveat to the results that are marginally significant (p-value very close to 5%). If standard errors were clustered at pilot level, then these results may have been reported as statistically insignificant.

The DiD specification facilitated aggregate level and further analysis. In each case, the analysis was conducted at the LSOA and month levels. For a given observation, the LSOA was either in the pilot group or not; and in the month of the observation, CHB had either begun, or not. The DiD considered these 2 variables only. As a result, new observations always functioned the same way, whether they were added by increasing the time period analysed or adding new pilot areas to the sample. The aggregate level and each type of further analyses all followed the same procedure, and the only difference was the sample of LSOAs considered. Therefore, no adjustment was required to account for different pilots starting CHB at different times.

Crime displacement: Bayesian time series (BTS)

There is a risk that the activities of CHB could result in the displacement of crime outside of the pre-determined pilot area into surrounding areas. If this occurred, then the local effectiveness of CHB would not reflect its overall impact. To assess crime displacement, a buffer was created around the pilot area depending on how urban the pilot area was, on a scale of 1 to 3 (3 is most urban: 3 to 0.5 miles, 2 to 1 mile, 1 to 1.5 miles). Crime data were analysed using BTS in the buffer zone to gauge whether there has been a crime displacement out of the pilot area.

The approach of BTS is to infer the effect the CHB intervention had on crime by analysing differences between expected and observed time series data. Data were divided into 2 parts: the first is known as the pre-intervention period and Bayesian updating was applied to fit a model that best explains what has been observed. The fitted model was used in the second part – the post-intervention period – to forecast what crime rates would look like had the intervention not taken place. The inferences are based on the observed crime to those predicted, which yields the absolute and relative expected effect of CHB on the data.

Technical methodology: Community survey

The community survey also adopted a quasi-experimental approach by comparing the survey results between residents of CHB areas versus residents of control areas. The control areas were the same as those identified in the impact evaluation on crime for the 6 areas surveyed.

Questionnaire development

There were 2 versions of the survey created (online and hard copy), and these were developed simultaneously. The primary aim was to ensure that both versions were sufficiently similar so responses gained via either version could be combined. An identical question set was included in both versions of the survey. The layout of questions in the online version of the survey also matched the hard-copy version.

The main difference between the 2 versions was the ability to enforce routing. Developing the survey using an online platform means that questions can be scripted so that a respondent is only presented with a particular question based on their answer to an earlier question(s). This was more difficult to enforce with the hard-copy version, which instead included written instructions for respondents to skip questions, dependent on their earlier answer(s). Where the online version includes routing by design, therefore eliminating any opportunity for respondents to answer questions not intended for them, the hard-copy version depended on a respondent reading and following the relevant instructions.

Data collection

There were 3 key stages to the data collection approach:

1) Push to online survey

The first stage was the physical delivery of an invitation to complete the survey online. These were disseminated using Royal Mail’s (RM) Door Drop Service, which was chosen as the most suitable approach given the disparate nature of some of the control sites, was more cost effective than direct mail to each address, and that it would allow for a larger sample to be reached than if dissemination of the survey was performed by the evaluation team or other fieldworkers. However, RM door drops can only be targeted down to postcode sectors (first part of postcode and first digit of second part) which meant many households around the sites also received invitations to the survey. This was seen as a potential benefit to the research, which would allow for an analysis of perceptions near CHB and control areas. Unfortunately, poor response rates made this analysis unviable. In addition, one CHB site was not covered by the RM’s Door Drop Service. Therefore, invitations were directly mailed to one postcode sector (comprising 300 addresses). Table 4 below outlines the total number of initial letters and follow-up postcards sent (including households outside of the pilot or control areas) and the response rates (excluding households outside of the pilot or control areas) for each sample.

2) Hard-copy survey

In addition to the initial invite letter, hard-copy surveys were directly mailed to addresses in one pilot area and its control areas 2 weeks after the initial invitation. This was to maximise the number of completions given the small size of the CHB area.

Hard-copy surveys were also directly mailed to respondents who requested a hard-copy version using the contact details provided on the initial invitation. This option was provided to minimise barriers to completion, such as digital poverty, and increase the reach of the survey. In total, 37 hard-copy surveys were received.

3) Postcard reminders

The final stage was the physical delivery of a postcard to households in all pilot and control sites, reminding them to complete the survey. This was also facilitated by RM’s Door Drop Service.

Table 4: Quantity of postcards and letters sent (5 pilot sites included in final analysis)

	Initial letters sent	Postcards sent	Households in sites	Responses received	Response rate
CHB sites	32,607	32,604	19,628	385	1.96%
Control sites	38,614	38,647	15,010	250	1.67%

Analytical methodology

The survey was open from 19 January 2024 until 2 April 2024. Upon closure, survey data was exported from the online survey platform to RStudio where it was cleaned and analysed.

A key part of the preparation for analysis was deciding how to count survey responses as being in a pilot or control site due to the RM Door Drop Service only available for whole postcode sectors (see above). This was based on the full postcode respondents provided at the start of the survey. Using this, a ‘strict’ definition and a ‘broad’ definition were established which represent 2 methods of assigning whether a response would be counted in a pilot or control site.

Both definitions are based on data from the Office for National Statistics (ONS) which is a best-fit lookup between postcodes, output areas (OAs) and lower layer super output areas (LSOAs) as defined by the 2021 Census. Postcodes are assigned to an OA by plotting the location of the postcode’s mean address in the areas of the output geographies. Each pilot or counterfactual site were defined by several OAs and/or LSOAs and now, using the best-fit lookup, the relevant postcodes for each site were matched.

The ‘strict’ definition is the product of that best-fit lookup. Each postcode is assigned to the OA/LSOA which houses most of the residents of that postcode, meaning that there are postcodes which touch on the area defined as a pilot or counterfactual site but are excluded from this definition. The ‘broad’ definition, however, includes any postcode that overlaps with a pilot or counterfactual site, regardless of how much.

Analysis of the survey was conducted using the ‘broad’ definition. This was decided as the ‘broad’ definition provided a greater sample size (635 versus 505) with a similar level of statistical power, and that most responses which would be excluded by the ‘strict’ definition, but included in the ‘broad’ definition, came from respondents who were in the immediate vicinity of the pilot or control site. Therefore, their experience of the local area and the subjects the survey asks about were expected to be sufficiently similar.

Once the pilot and control sample had been finalised, respondents were weighted based on key criteria to mitigate against the lack of representativeness of survey results. These criteria were:

Site weighting: Individual sites were weighted to ensure that each contributed a similar proportion to the entire sample. This approach allowed a balance between ensuring that one or 2 sites do not overly dictate the shape of the results, while retaining the original differences and order between the sites.

Demographic weighting and balancing of pilot and control areas: The distribution of gender, age, and ethnicity in the sample was matched to the distribution of those characteristics in the population at an aggregate level across all pilot and control areas. The original categories for gender (male and female) were retained, however, categories for age and ethnicity were aggregated to create feasible weighting criteria. To perform this weighting, relevant population data relating to gender, age and ethnicity from the 2021 Census were collated for each of the LSOAs and/or OAs which constituted a pilot or control area. This ensured the pilot and control sample matched each other as much as possible and thus maximised the likelihood that any difference in results between the 2 groups was due to CHB and not demographic differences.

The weighting approach resulted in an effective sample size of 475, and a margin of error around estimates of +/- 4.5. The minimum and maximum weight applied to a respondent are outlined in Table 5 below.

Table 5: Minimum and maximum weights applied

Minimum	1st quartile	Median	3rd quartile	Maximum
0.3557	0.5753	0.8054	1.3432	5.6033

The approach taken to weight the survey data based on the above criteria was chosen as it provided the most suitable balance between ensuring that the pilot and control samples:

• matched each other as much as possible and thus maximised the likelihood that any difference in results between the 2 groups was due to CHB and not demographic differences
• matched the demographics of the areas they represent and thus are a robust reflection of the wider population
• did not overly-inflate the responses of a few individuals by giving them substantially higher weightings than other survey respondents or unbalance the sample, which would have reduced the effective base size

Before agreeing on the final approach, several weighting options were considered, each offering a different trade-off between those 3 points.

There were several factors that ideally would have been incorporated in the data weighting to minimise the likelihood that the results were a product of different demographics across the pilot and control samples. These included age, gender, ethnicity, and the balance of distribution of the samples across each pilot and control site.

It was agreed that the samples should reflect the perceptions across all 5 sites as equally as possible, and so the proportional contribution of each area was confirmed early on. It was not possible to construct a weighting structure that matched each pilot and control area to its population proportions for age, gender, and ethnicity. While the gender profile within each site sample was relatively close to the population, as discussed earlier, the age profile skewed substantially older, making a weighting scheme that properly represented 16 to 24 years unviable. It was also not possible to weight the ethnicity data to represent the local populations due to most areas being substantially skewed (over 80%) towards white respondents, meaning that other respondents would have to be significantly up weighted. This resulted in the decision to weight both age and ethnicity to aggregated population proportions, as the analysis was being conducted at an aggregate level.

This still required the grouping of age and ethnicity categories to 2 categories each, to allow a weighting solution to be found when combined with the additional criterion of balancing the contribution of each site to the overall aggregate pilot and control samples. The purpose of this was to ensure areas from one police force did not contribute, for example, a significant proportion to the pilot site sample and very little to the control site sample and therefore potentially introduce confounding variance into the data due to an area having substantially different perceptions than the others.

This weighting scheme was selected based on the balance of trade-offs. While the maximum weighting of 5.6 is recognised as undesirable, it was only applied to one individual in the sample, and very few respondents receive a weighting above 3. This was accepted as the best available option, balancing the needs outlined above. Table 6 below shows the final weighting targets applied to each group.

Table 6: Combined aggregate totals of gender, age and ethnicity and resulting weighting targets

	Demographic	% of population	% of total pilot or control in the sample	Weighting target (%)
Pilot			61%
	Male	49%		30%
	Female	51%		31%
Control			39%
	Male	49%		19%
	Female	51%		20%
Pilot			61%
	44 or under	55%		33%
	45 and over	45%		28%
Control			39%
	44 or under	54%		21%
	45 and over	46%		18%
Pilot			61%
	White	65%		21%
	Other	35%		39%
Control			39%
	White	71%		12%
	Other	30%		28%

Annex B: Full crime impact results

Table 7: Sensitivity analysis results – Reasonable worst- and best-case scenarios

Category	Crime grouping	Effect size (crimes per 1,000 residents, per month)	p-value (significant values marked with *)
Worst case (2 best performing sites removed from sample)	Total PRC	0.50	0.552
	ASB proxy	0.17	0.449
	Acquisitive crime	-0.29	0.057
	Drugs	-0.07	0.659
	Violence	0.08	0.791
Best case (2 worst performing sites removed from sample)	Total PRC	-1.91	0.006*
	ASB proxy	-0.51	0.011*
	Acquisitive crime	-0.37	0.002*
	Drugs	-0.28	0.022*
	Violence	-0.65	0.010*

Table 8: Further analysis results – CHB launch

Category	Crime grouping	Effect size (crimes per 1,000 residents, per month)	p-value (significant values marked with *)
Crime and harm mapping	Total PRC	-3.7	0.001*
	ASB proxy	-0.87	0.001*
	Acquisitive crime	-0.32	0.002*
	Drugs	-0.17	0.045*
	Violence	-1.47	0.000*
Deprivation	Total PRC	1.87	0.168
	ASB proxy	0.64	0.092
	Acquisitive crime	-0.32	0.215
	Drugs	0.46	0.055
	Violence	0.59	0.192
Reaction	Total PRC	-0.74	0.649
	ASB proxy	-0.57	0.315
	Acquisitive crime	-0.57	0.216
	Drugs	-0.54	0.311
	Violence	0.16	0.85

Table 9: Further analysis results – Progress

Category	Crime grouping	Effect size (crimes per 1,000 residents, per month)	p-value (significant values marked with *)
Clear only	Total PRC	0.07	0.982
	ASB proxy	-0.35	0.653
	Acquisitive crime	-0.46	0.412
	Drugs	-0.15	0.818
	Violence	0.67	0.585
Clear and Hold for 6+ months	Total PRC	-0.97	0.289
	ASB proxy	-0.38	0.134
	Acquisitive crime	-0.41	0.014*
	Drugs	0.13	0.402
	Violence	-0.33	0.256
Clear, Hold for 6+ months, and Build for 6+ months	Total PRC	-0.94	0.127
	ASB proxy	-0.05	0.821
	Acquisitive crime	-0.26	0.011*
	Drugs	-0.09	0.215
	Violence	-0.4	0.039*

Table 10: Further analysis results – Area type

Category	Crime grouping	Effect size (crimes per 1,000 residents, per month)	p-value (significant values marked with *)
Urban	Total PRC	-1.86	0.075
	ASB proxy	-0.25	0.412
	Acquisitive crime	-0.43	0.016*
	Drugs	-0.39	0.050*
	Violence	-0.92	0.024*
Rural	Total PRC	-1.04	0.259
	ASB proxy	-0.22	0.489
	Acquisitive crime	-0.3	0.091
	Drugs	-0.05	0.602
	Violence	-0.03	0.882
Small	Total PRC	-0.69	0.805
	ASB proxy	0.61	0.584
	Acquisitive crime	-0.78	0.321
	Drugs	-0.47	0.611
	Violence	-0.22	0.885
Medium	Total PRC	-3.85	0.000*
	ASB proxy	-0.91	0.001*
	Acquisitive crime	-0.5	0.000*
	Drugs	-0.2	0.028*
	Violence	-1.43	0.000*
Large	Total PRC	0.57	0.257
	ASB proxy	0.13	0.521
	Acquisitive crime	0.04	0.659
	Drugs	-0.02	0.685
	Violence	0.35	0.037*

Annex C: Full community survey question set and results

See the community survey question set and results here:

Annex C: Community survey results

Annex D: Monitoring data

Data collection and analysis

Data for some outcome metrics was collected directly from police forces as it is not required to supply this data to the HO. Forces were asked to provide monthly data on the following metrics for the 12-month period before and each month after CHB was implemented, outlined below.

1) MoRiLE scores

There are 2 types of MoRiLE scores which attempt to provide a consistent approach to identifying tactical and strategic policing priorities across law enforcement agencies:

• Tactical MoRiLE scores in the context of CHB calculate the harm of an OCG according to several different factors, such as their physical or psychological harm, or geographic scope (for more information, see College of Policing (2020))
• Thematic MoRiLE scores focus on the vulnerabilities associated with the strategic issue being addressed (which in the case of CHB, are SOC hotspots) and assess potential impact and harm on the individual, community, and environment

A higher MoRiLE score indicates a higher level of harm and threat; pilot sites were asked to provide a Tactical MoRiLE scores for each OCG mapped to the area, in addition to a Thematic MoRiLE for their CHB area.

2) Community intelligence reports:^{[footnote 14]} pilot sites were asked to provide data on the number of intelligence reports related to the CHB area where the source of information was a member of the public

This data aimed to indicate trends in these outcomes through a pre-post CHB comparison, giving additional context to the crime impact evaluation results. The level of completeness varied between sites. For example, some sites could not provide data on community intelligence reports for the pre-CHB period due to capacity constraints. In others, an OCG had not been MoRiLE scored before CHB for a pre-post comparison. This limited comparability between sites. It is also important to highlight that, because the equivalent data was not collected for control areas, any changes in these outcomes cannot confidently be attributed to CHB. Therefore, these metrics are not an objective measure of the impact of CHB and should be interpreted with caution.

Results

Nine pilot sites returned data on Tactical MoRiLE scores for OCGs mapped to the CHB area. Across these sites, scores for 21 OCGs were provided. However, only 5 sites provided a Thematic MoRiLE score for the CHB area. Interviews suggested this may be due to a lack of guidance and understanding of Thematic MoRiLE scores. Informal conversations with forces highlighted this may also be due to limited analytical resource. Table 11 shows that, overall, there has been a decrease in Tactical and Thematic MoRiLE scores over the course of CHB implementation. However, as highlighted above, these findings should not be viewed as evidence of impact due to the absence of control area scores and the subjective nature of MoRiLE scoring.

Table 11: Median change in MoRiLE scores

	Median raw change	Median percentage change
Tactical MoRiLE	-95	-52%
Thematic MoRiLE	-162	-75%

Notes:

1. Tactical MoRiLE: Presents the median Tactical MoRiLE change of 21 OCG scores. For 6 Tactical scores, the change was calculated based on the latest score pre- and post-CHB. The remaining 15 did not have pre-CHB MoRiLE scores. In these cases, the analysis compared the first and most recent score post-CHB.
2. Thematic MoRiLE: Presents the median Thematic MoRiLE score change of 5 area-based scores. In all cases, the analysis compared the first and most recent score post-CHB.

Annex E: Proforma survey results

Table 12: Number of sites for each SOC threat, by priority ranking

	High	Medium	Low	Not relevant
Illegal drugs supply	11	2	0	0
Illegal firearms	5	4	1	3
Modern slavery and human trafficking	5	1	3	4
Other	5	0	0	8
Organised acquisitive crime	4	2	5	2
Child sexual exploitation and abuse	4	3	3	3
Money laundering and other economic crime	2	7	3	1
Fraud	1	2	4	6
Organised immigration crime	1	2	3	7
Bribery and corruption	0	4	5	4
Cyber crime	0	1	4	8
Base size: 13 sites

Table 13: Number of sites for each crime type, by priority ranking

	High	Medium	Low	Not relevant
Drug trafficking	10	3	0	0
Drug possession offences	9	3	1	0
ASB	8	3	2	0
Possession of weapons	7	5	1	0
Violence with injury	6	6	1	0
Violence without injury	6	3	4	0
Public order	5	3	3	2
Knife crime offences	4	6	2	1
Firearms discharges	4	0	5	4
Threats to life	3	4	4	2
Homicide	3	1	4	5
Theft of vehicle	2	2	6	3
Robbery	2	4	3	4
Sexual offences	2	2	4	5
Rape	2	2	3	6
Criminal damage (excluding arson)	2	3	7	1
Burglary (residential)	1	1	9	2
Other	1	0	0	12
Burglary (business)	0	2	7	4
Stalking and harassment	0	4	6	3
Malicious comms	0	0	8	5
Misc. crimes against society	0	2	6	5
Arson	0	3	8	2
Base size: 13 sites

Table 14: Delivery progress of sites, by phases started or completed (excluding sites which provided incomplete information)

	Number of sites
Clear	2
Clear and Hold	2
Clear, Hold, and Build	7
Base size: 11 sites

Table 15: Approach to hotspot selection

	Number of sites
Only pre-existing knowledge	2
Pre-existing knowledge and new analysis	9
New analysis only	0
Other	2
Base size: 13 sites

Table 16: Data sources used to inform hotspot selection (sites which used new analysis to inform hotspot selection)

	Number of sites
PRC data	8
MoRiLE scores	6
Crime severity scores	5
Informal evidence from police officers / policing leads	8
Community surveys	3
Anecdotal evidence from community	9
Data / intelligence from wider agencies	9
Other	1
Base size: 9 sites

Table 17: Scope of locality profile created as part of CHB planning (sites which created a locality profile as part of their CHB planning)

	Number of sites
Profile of wider force area created to identify a CHB area only	1
Locality profile created of CHB area after identifying site only	5
Both types of locality profile created	2
Base size: 8 sites

Annex F: Online stakeholder survey results

See full results here:

Annex F: Online stakeholder survey results

Annex G: Programme-level theory of change

The programme-level ToC was developed through collating information from existing resources on CHB (such as HO practitioner guidance) and direct engagement with pilot sites via the proforma survey.

See the ToC here:

Annex G: CHB programme-level theory of change

Annex H: Site list

Table 18: Evaluation pilot sites

Police force area	CHB area
Bedfordshire	Dallow
Cambridgeshire	Lincoln Road Corridor
Cheshire	Murdishaw
Durham	Horden
Greater Manchester	Cheetham Hill
Merseyside	Birkenhead
Merseyside	Dovecot / Knowsley
Merseyside	Sefton
Metropolitan Police	Barnet
Northumbria	Benwell
Northumbria	Easington Lane
South Yorkshire	Hoyland
West Yorkshire	Bradford Moor

References

Abadie, A. & Gardeazabal, J. (2003) ‘The Economic Costs of Conflict: A Case Study of the Basque Country’, American Economic Review, vol. 93(1), pp. 113-132 [accessed 29/07/24]

Abadie, A., Diamond, A. & Hainmueller, J. (2010) ‘Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California’s Tobacco Control Program’, Journal of the American Statistical Association, vol. 105(490), pp. 493-505 [accessed 29/07/24]

Bradburn, M.N., Rips, L.J., & Shevell, S.K. (1987) ‘Answering Autobiographical Questions: The Impact of Memory and Inference on Surveys’, Science, vol. 236(4798), pp.157-161; [accessed 25/07/24]

Campana, P. & Varese, F. (2018) ‘Organized Crime in the United Kingdom: Illegal Governance of Markets and Communities’, The British Journal of Criminology, vol. 58(6), pp. 1381-1400; [accessed 29/08/2024]

College of Policing (2021) Protective measures and civil orders [accessed 04/11/2024]

College of Policing (2020) Evaluation and review [accessed 04/11/2024]

College of Policing (2015) Intelligence report [accessed 04/11/2024]

College of Policing (2013) Strategic planning [accessed 04/11/2024]

Coomber, R. (2015) ‘A tale of two cities: Understanding differences in levels of heroin/crack market-related violence – a two cite comparison’, Criminal Justice Review, vol. 40(1), pp. 7-31; [accessed 05/11/2024]

Crocker, R., Webbs, S., Skidmore, M., Garner, S., Gill, M., & Graham, J. (2019) ‘Tackling local organised crime groups: lessons from research in two UK cities’, Trends in Organized Crime, vol. 22, pp. 433-449; [accessed 05/11/2024]

Crocker, R., Webb, S., Garner, S., Skidmore, M., Gill, M. & Graham, J. (2017) The Impact of Organised Crime in Local Communities [accessed 05/11/2024]

Hales, G. & Hobbs, D. (2010) ‘Drug markets in the community: a London borough case study’, Trends in Organized Crime, vol. 13, pp. 13-30; [accessed 05/11/2024]

HMICFRS (2022) ‘About the data’. His Majesty’s Inspectorate of Constabulary and Fire & Rescue Services [accessed 04/11/2024]

Home Office (2023a) No Place to Hide: Serious and Organised Crime Strategy 2023-2028 [accessed 05/11/24]

Home Office (2023b) Criminal exploitation of children and vulnerable adults: county lines [accessed 04/11/2024]

Home Office (2024a) Evaluation of the Safer Streets Fund round 3, year ending March 2022 [accessed 04/11/2024]

Home Office (2024b) Safety of Women at Night Fund Evaluation [accessed 05/11/2024]

Home Office (2024c) Evaluation report on Grip and bespoke-funded hot spot policing [accessed 04/11/2024]

Kibuacha, F. (2024) Survey response rates: A guide to higher survey completions [accessed 25/07/2024]

Kirby, S., McManus, M. & Boulton, L. (2018) ‘Examining the demographic profile and attitudes of citizens, in areas where organized crime groups proliferate’, Trends in Organized Crime, vol. 21, pp.172-188; [accessed 05/11/2024]

Koper, C.S., Taylor, B.G., Liu, W. & Wu, X. (2022) ‘Police activities and community views of police in crime hot spots’, Justice Quarterly, vol. 39(7), pp. 1400-1427 [accessed 05/11/2024]

Lavorgna, A., Lombardo, R. & Sergi, A. (2013) ‘Organized crime in three regions: comparing the Veneto, Liverpool, and Chicago’, Trends in Organized Crime, vol. 16, pp. 265-285 [accessed 05/11/2024]

Lupton, R., Wilson, A., May, T., Warburton, H. & Turnbull, P.J. (2002) A rock and a hard place: drug markets in deprived neighbourhoods, Home Office Research Study 240 [accessed 05/11/2024]

NCA (nd) County Lines. National Crime Agency [accessed 04/11/2024]

Pitts, J., Hope, T., Hurley, M. & McGibbon, I. (2017) Preventing Organised Crime [accessed 05/11/2024]

Rosenbaum, P.R. & Rubin, D.B. (1983) ‘The Central Role of the Propensity Score in Observational Studies for Causal Effects’, Biometrika, vol. 70(1), pp. 41-55; [accessed 05/11/2024]

Ucko, D.H. (2013) ‘Beyond Clear-Hold-Build: Rethinking local-level counterinsurgency after Afghanistan’, Contemporary Security Policy, vol. 34(3), pp. 526-551; [accessed 05/11/2024]

Vilalta-Perdomo, C., Lopez-Ramirez, P. & Fondevila, G. (2021) ‘A Community-level Test of General Strain Theory (GST) in Mexico’, Deviant Behaviour, vol. 43(11), pp. 1331-1346; [accessed 05/11/2024]