Hot spot policing in England and Wales, year ending March 2023: Evaluation of Grip and bespoke-funded hot spot policing
Published 27 March 2025
Applies to England and Wales
Executive summary
Background
Following 2 successful pilots, the Home Office (HO) announced in April 2021 that 18 police forces with the highest levels of serious violence (SV) would receive funding to deliver enhanced hot spot policing. The aim of this programme, called Grip, was to deter SV through visible patrol activity in hot spots while also adopting strategic problem-oriented policing (POP) to address the root causes of violence within those locations. In September 2021, 2 further police forces were awarded bespoke funding to conduct hot spot policing, as they had the next highest volumes of SV. The 20 forces had a single-year Grant Agreement for the year ending 31 March 2022 and then a multi-year agreement for the next 3 years (though see below), to deliver the hot spot policing programme. Following consultation with leading hot spot policing scholars, we believe this is the first attempt to implement a national hot spot strategy and evaluate it robustly.
A separate Innovation Fund was made available in the year ending 31 March 2023. Police forces had the option to bid for additional funding to test new, innovative approaches to hot spot policing that would not have been delivered as part of their core Grip activity. Five police forces were successful in bids to receive Innovation Funding.
For the year ending March 2025, the Grip grant was replaced by the Hot Spot Response programme (a single-year grant), which expanded the programme to include all 43 territorial police forces in England and Wales and to focus on reducing actual and perceived antisocial behaviour (ASB), as well as SV.
This evaluation is the second in a series of reports measuring the impact of the hot spot funding. The first report, Evaluation report on Grip and bespoke-funded hot spot policing, covers the year ending 31 March 2022 (Home Office, 2024). It estimated that Grip and the bespoke hot spot funding resulted in a 7% reduction in violence and robbery offences in the hot spots on days patrolled versus days not patrolled. This second report mainly contains results for the year ending March 2023, but it also includes some new results that were generated for the previous year.
Aims and methodology
This report details HO evaluation of the Grip and bespoke hot spot policing funds for the financial year ending March 2023. The aims of the evaluation are to:
-
provide an assessment of how much hot spot policing is being delivered in England and Wales and how it is being delivered
-
evaluate the effect of visible patrolling on the crime rate within a hot spot area – our main results test violence and robbery offences but this report also includes results for other types of crime for the year ending March 2023 and the year prior
-
evaluate the effect of visible patrolling on the volume and rate of positive crime outcomes within a hot spot area for the year ending March 2023 and the year prior – positive outcomes are a measure of how many crimes are solved or cleared
-
evaluate whether visible patrolling in hot spots causes crime to move to adjacent areas (displacement) or whether the patrols reduce crime in those areas (diffusion of benefit)
-
assess the cost effectiveness of Grip and bespoke-funded hot spot policing
-
provide forces with regular feedback to allow continuous improvements to be made to their hot spot policing approach
-
expand the evidence base on hot spot policing, particularly in the UK
No police forces’ approach to hot spot policing were identical, and some forces ran multiple different programmes. Therefore, this report effectively comprises 22 different evaluations from 19 forces, 11 of which randomised their patrolling in some way to improve the robustness of the impact estimates. We were unable to evaluate the effects of hot spot policing in one force, due to data issues. Most results were generated using a repeat crossover design, which compares crime on days patrolled against days with no patrols. These were brought together in a meta-analysis to assess the overall national effect. The meta-analysis includes 18 model results from 18 forces. That left 4 results reported separately. Two used an alternative approach in which effects were generated by comparison with matched control areas, (one of these was also an Innovation Fund result) and the final 2 were Innovation Fund results reported separately from the main analysis.
Key Findings
At least 90,000 patrols were carried out by the forces receiving bespoke hot spot policing funding in the year ending March 2023 and over 85,000 weapons were collected.
Results for the meta-analysis did not suggest that the programme had a statistically significant impact on volumes of violence and robbery offences in the hot spots in the year ending March 2023 (the exact basket of offences used is in Annex G). This differs from the 7% reduction reported for the national programme in the year ending March 2022 (Home Office, 2024) and the even larger crime reduction effects found in the pilots (Basford and others, 2021; Bland and others, 2021). However, there was a statistically significant violence/robbery reduction for one of the Innovation Fund projects, which involved using uniformed and plain-clothes patrols and a social media campaign.
There may be several reasons for the lack of programme-level impact. It is possible that hot spot policing was not implemented as well in the year ending March 2023 as in previous years, or that hot spot policing effects wane over time. While the existing evidence base for hot spot policing is very strong, most studies do not test the effects for longer than a year. Another possibility is that patrolling became less effective because forces put more of their efforts and spent more of their funding on POP in the year ending March 2023. We have not evaluated the effectiveness of POP activity in this report, but will include it in the next update. It is also possible that our methodology may become less effective at detecting an effect over time.
Our model compares days patrolled versus days not patrolled. Over time, it is possible that the accumulation of patrols causes a general deterrence effect that lowers crime on all days, regardless of whether there is a patrol or not. Our judgment is that waning or uneven implementation is the most likely reason, but further analysis is needed to cement this conclusion.
This report also estimated the effect for other types of crime. Generally, we did not find significant impacts in either year for other types of crime. In one sense, this is not surprising. The hot spots were selected because of their high levels of serious violence, not because of their levels of other crimes. However, it is worth noting that many other hot spot patrol studies have found effects on a wide variety of crime types.
Results did not suggest any displacement of crime to nearby areas, nor any diffusion of benefits. This was tested using a 50-metre buffer area for a subset of the year ending 2022 results, in which we found an 11% reduction in violence/robbery in the hot spots themselves. It was not tested for the year ending March 2023 because there was no significant effect in the hot spots. The lack of evidence for displacement is consistent with other findings in the literature.
While this study did not find an impact on volumes of crimes in the year ending March 2023, it did find an impact on crime outcomes, which is a measure of crimes solved and a current government priority. Positive outcomes for violence/robbery (a measure of the number of crimes solved or cleared, explained in detail in Chapter 1 and Annex G) increased by 8.1% in hot spots on days patrolled versus days not patrolled. Similar impacts were observed in other crime baskets, including all crimes minus possession, victim-based crimes, and possession, with significant increases of 7.8%, 8.2% and 20.0% on patrol days, respectively. This suggests the programme is resulting in more robbery and violence offences being solved and more weapons and drugs being found in the hot spots. Similar results were found for the year ending March 2022. Reasons why this might be the case are explored in Chapter 3.
Cost-benefit analysis revealed that the social cost of the elements of the programme evaluated in this report for the year ending March 2023 was £17.5 million in financial year to March 2025 prices. It was not possible to monetise any benefits for the programme in that year, because there was no significant impact on the number of offences at the programme level. There was an impact on positive outcomes, indicating that more crimes were solved in the hot spots due to the programme, but the Home Office does not yet have a method for accurately monetising those impacts. This means we cannot say that the programme was cost effective in the year ending March 2023. However, it is worth noting that the total crime reduction benefits in the year ending March 2022 were estimated at £38.4 million with programme costs totalling £16.8 million in financial year to March 2025 prices. That means overall, the programme’s crime reduction benefits (£38.4 million) exceed the total cost across both years (£34.3 million).
Acknowledgements
The authors would like to thank several colleagues for their support with the research. Particular thanks go to Ricky Wang and Tom Kirchmaier at the London School of Economics for their continued support and technical guidance. Thanks to the funded police forces who facilitated the evaluation, as well as the force analysts and academics who provided their own internal evaluation results. Thanks also to the Grip policy team (Jack Duncton, Sarah Kalmus-Hoye, Aoife Genoni, Naomi Cooper and Libby Wells), Peppa Pancheva and Chris Linehan in the Crime Statistics team, and to Peter Neyroud for his support with peer review of the report. Finally, the authors would like to thank University College London (Spencer Chainey and Aiden Sidebottom) and the College of Policing for providing forces with their expertise through individual force assessments and for their support in delivering hot spot policing.
1. Introduction, aims and methodology
1.1 Policy background
In 2019, the Home Office (HO) announced that 18 police force areas (PFAs) would receive funding for the year ending 31 March 2020 to enhance the operational policing response to SV. HO selected the areas by their levels of hospital admissions for injury with a sharp object experienced between the years ending 31 March 2016 and 2018. In early 2020, HO confirmed funding for this approach, known as Surge, for a second year (April 2020 to March 2021). Within Surge, PFAs were provided with broad discretion to apply the funding in the manner they considered would be most effective, provided it was aimed at interventions to reduce SV, particularly homicide, knife crime and gun crime.
During the year ending 31 March 2021, 2 police forces used Surge funding to run hot spot policing pilots. The results were highly successful – see below. From April 2021, Grip funding replaced Surge and required the 18 forces to deliver enhanced hot spot policing activity. Grip aimed to immediately suppress and reduce SV through visible patrol activity in hot spots while also adopting strategic POPto address the root causes of violence in hot spots over the long term.
The same local areas which received Grip funding also received funding for the development of Violence Reduction Units (VRUs), which, like Grip, aim to reduce SV but focus on multi-agency prevention programmes.
In September 2021, Cleveland and Humberside were awarded bespoke funding to address increases in SV because they were the next 2 highest forces for hospital admissions for injury with a sharp object. The purpose of the funding was to implement hot spot policing in the highest SV hot spots in Cleveland and Humberside. As their approaches were similar to the Grip-funded forces, HO included them in the analysis for the evaluation of hot spot policing.
The 20 forces had multi-year Grant Agreements (for the years ending 31 March 2022 to 31 March 2024) to deliver the hot spot policing programme. As far as could be determined, Grip was one of the largest attempts to implement and measure hot spot policing ever. A separate Innovation Fund was made available in year ending March 2023. Police forces had the option to bid for additional funding to test new, innovative approaches to hot spot policing that would not have been delivered as part of their core Grip activity. Five police forces were successful in bids to receive Innovation Funding. Successful bids ranged from working with partner agencies to deliver additional patrolling to using drone technology to increase visibility and CCTV capabilities. Further details of the activity conducted by each police force under this fund are available in Annex E.
In the year ending March 2025, the programme was broadened to include all 43 territorial police forces in England and Wales, and to focus on reducing actual and perceived ASB, as well as SV.
1.2 Hot spot policing
Hot spot policing is a place-based policing intervention that focuses police resources and activities on those places where crime is most concentrated. For Grip, hot spots constitute small geographical areas, often specific streets or neighbourhoods and generally not larger than a Lower Super Output Area (LSOA), which is a unit of geography comprising between 400 and 1,200 households with a resident population between 1,000 and 3,000 persons. These are identified using data and intelligence as experiencing the highest volumes of SV. Focusing resources and activities in hot spots aims to prevent crime in these specific areas and potentially reduce overall crime levels in the wider geographic area.
There is an extensive evidence base, both in the UK and internationally, demonstrating the effectiveness of hot spot policing. Key findings from this research are summarised below.
Studies have shown that proactive, visible police activity in hot spots can decrease crime and disorder (Weisburd and others, 2008). In a meta-analysis of 65 studies, Braga and others (2019) found that hot spot policing programmes produced statistically significant (p < 0.05) positive mean effect sizes for violent crime outcomes (0.102), property crime outcomes (0.124), disorder outcomes (0.161) and drug crime outcomes (0.244).
Some tactics work better than others: problem-solving at hot spots seems particularly effective at reducing crime and decreases crime for longer than visible patrols (Taylor and others, 2014; Braga and others, 2019).
Recent UK-based evidence of effectiveness of visible patrols comes from 2 Surge-funded pilots. Operation Ark (Basford and others, 2021) in Essex delivered high visibility patrols for 15 to 20 minutes at 20 high-harm, square-shaped hot spots of 150m by 150m. Violent crime, visible street crime harm, community violence and its related harm were all reduced in patrolled areas on patrolled days compared to non-patrolled days. The second Surge-funded pilot was also a randomised controlled trial (RCT) in Bedfordshire. It found that even minimal amounts of foot patrol can prevent SV across a large area. The study also found evidence of a cumulative effect on crime, with the largest reduction in crime harm found after 3 days of consecutive patrol in the same LSOA (which are units of geography comprising between 400 and 1,200 households and usually have a resident population between 1,000 and 3,000 persons.). However, this was with a small sample size of 21 hot spots over 90 days with pandemic restrictions in place (Bland and others, 2021).
1.3 Evaluation aims and objectives
The aims of this work are to:
-
provide an assessment of how much hot spot policing is being delivered in England and Wales and how it is being delivered, including whether patrols are randomised and/or tracked using GPS technology
-
evaluate the effect of visible patrolling on the crime rate within a hot spot area – our main results test violence and robbery offences, but this report also includes results for other types of crime for the year ending March 2023 and the previous year
-
evaluate the effect of visible patrolling on the volume and rate of positive crime outcomes within a hot spot area for the year ending March 2023 and the previous year (see Annex G for a list of the crime outcomes included in this basket)
-
evaluate whether visible patrolling in hot spots causes crime to move to adjacent areas (displacement) or whether the patrols reduce crime in those areas (diffusion of benefit)
-
assess the cost effectiveness of Grip and bespoke-funded hot spot policing
-
provide forces with regular feedback to allow continuous improvements to be made to their hot spot policing approach
-
expand the evidence base on hot spot policing, particularly in the UK
Much of the existing research on hot spot policing has been conducted in the US, which has different levels and types of crime and potentially different crime drivers. There have been fewer UK studies that have tested the effect of hot spot policing on violence and SV and most of those have tended to be small-scale pilots. Grip presents an opportunity to test the effects of hot spot policing on these crime types on a much larger scale and over a long period.
Originally, HO hoped to test the effect of the hot spot patrols on the most serious categories of violence, like homicide and knife/firearm-enabled crime. However, numbers within these offence groupings were too small for meaningful evaluation analysis. Instead, police-recorded robbery and violence against the person were used. Offences intrinsically linked to the presence of officers and officer action in a location were removed. By including violence against the person and robbery, homicides and a large subset of the knife and firearm-enabled offences were also captured. See Annex G for the full list of included offence categories.
1.4 Evaluation methodology
The different elements of the evaluation methodology are:
Delivery metrics: Force patrol data and information were processed to allow for an assessment of how hot spot policing is being delivered and how much is being delivered.
Impact evaluation: A meta-analysis containing 18 hot spot policing evaluations with a repeated crossover design, including 10 with randomisation by day of patrol. Two additional results that compare treatment and control areas (one of which is also an Innovation Fund result) and 2 other Innovation Fund results. Displacement/diffusion effects were also tested for results that had significant impacts in the hot spots.
Cost-benefit analysis: Used to estimate the societal costs and benefits of the funded hot spot policing programmes in the year ending March 2023.
1.4.1 Data and information used
HO used several sources of data, as described below. The first 3 were bespoke data collections from forces, set up as part of the hot spot funding conditions. The fourth data source – the crime data – was part of a separate data stream between forces and HO that pre-dates the funding.
Force patrol activity data by date and hot spot
Forces provided data on their patrol activity, tracked through a variety of methods (such as GPS trackers, officer-completed template returns). The geometry and size of the treatment areas were largely at the discretion of the police force. Consequently, there was some variation across forces in the implementation of the Grip programme. In some, units were defined with free-form shapefiles, others used hexagons or boxes and some used patrol beats or administrative boundaries.
Data on weapons collected during the funded period
Forces provided a data return giving aggregated totals of weapons collected by quarter. Forces varied in their IT systems and recording mechanisms for weapons data. This included the collection method (such as seized, surrendered, collected through a weapon amnesty), and type of weapon (knife/sharp instrument, firearm, other). Forces also varied in whether they could identify those solely collected on funded hot spot patrols.
Financial information
Forces also provided data on the proportion of their allocated budgets that went on analytical systems, analytical capability, operational policing and problem-solving. Crime and outcomes data by date at the same geography as patrol data: Forces are required to provide crime data quarterly to the Home Office Data Hub under the Annual Data Requirement (Home Office, 2022). For this evaluation, crime locations were mapped and aggregated to the area level matching the activity data.
Social cost of crime estimates
The cost-benefit section of this report uses published figures on the social costs of different offences (Heeks and others, 2018).
1.4.2 Analytical approach
Delivery metrics
The data received by forces was cleaned using code developed in R, a programming language for statistical computing and graphics. This was to standardise date formats, correct typographical errors in area names, filter for relevant activity, and so on. Obvious errors were also removed, and further refinement sometimes took place following discussions with forces. R code was then used to pull out specific metrics from the data, notably the number of patrols and the days on which they took place. HO manually checked this process in a sample of forces.
Impact evaluation
No police forces’ approach to hot spot policing were identical. Therefore, this report effectively comprises 22 different evaluations. HO analysed 18 by using a repeat crossover design that compared crime rates in hot spots on days with patrols (‘days on’) and days without patrols (‘days off’). In this type of day-on-day-off model (DODO model), each area serves as its own control. In 10 cases, the mixture of days on and days off was randomised. However, due to operational practicalities and compliance, actual visitations were often non-random. Certain days of the week had lower compliance, for example. For that reason, a full set of control variables (see below) were used regardless of whether the force randomised or not.
For non-randomised approaches, police may actively try to patrol during high-crime days or more frequently during high-crime periods. To account for this, the regression model introduces controls to account for the area’s crime rate, within-week variations and seasonality in crimes. In the previous report, results were shown for both a negative binomial regression model and a Poisson regression model. Filters were also applied to remove areas with very low crime or very low patrol dosage. Following academic guidance and advice, those filters were dropped and this report uses the Poisson model for all results, which are reported with clustered robust standard errors. As the previous report showed, results from the 2 models are generally very similar. Annex A discusses the regression model in more detail.
Although this report generally covers performance of the programme in the year ending 31 March 2023, timings at the individual force level often did not match to financial years. Many forces revised their hot spots at the start of a new financial year but took a month or 2 to operationalise this. In practice, this meant the old hot spots continued to be patrolled for a portion of the new financial year. The previous report used a strict cut-off of 31 March 2021; however, it made little sense to take the same approach this time. That would have meant several forces would have had 2 operations to measure: one very short period with the old hot spots and a longer period with new hot spots. For that reason, we decided to relax the financial year boundaries and to let groupings of hot spots dictate the exact time period used. This means the exact time periods measured differ by force. It also means results for the year ending March 2022 were re-run with slightly altered time periods so that no data was excluded. Further details of this process and those results can be found in Annex B.
HO applied the regression model to each force’s operation, which yielded an estimate and confidence intervals (CI) for the treatment effect at the force level. A pooled effect for the programme as a whole (meta-analysis) was calculated by taking the average of the individual force-experiment estimates. This comprised results from 18 forces. Two-tailed 95% CI were applied to both the force-level results and the meta-analysis. The meta-analysis used was unweighted, meaning every force’s result was counted equally. Some studies alternatively use a weighted meta-analysis. For sensitivity, a weighted result was calculated using an inverse variance-weighted average model with random effects (see Annex D).
This method was used to estimate the impact on the different baskets of crime and positive outcome volumes/rates. We tested both volumes and rates in order to fully understand the impacts of the programme. It is important to understand whether volumes of positive outcomes are increasing because this impacts on the rest of the Criminal Justice System. More people are charged with an offence, meaning more court and possibly prison time are required. But because hot spot policing can also change the level of offences, changes in positive outcomes do not necessarily indicate a change in the rate at which crimes are being solved, which is an important metric for deterrence and crime reduction. For example, if offences increase by a similar amount to positive outcomes, the positive outcome rate (or crime clearance rate) might stay more or less the same. That is why we tested both volumes and rates. A full list of the outcomes included in our positive outcomes basket can be found in Annex G.
The baskets of crimes tested are violence and robbery, ASB-related offences (not incidents), possession (of weapons and drugs), neighbourhood crime, and all crimes without possession. A full list of offences for each basket of crime can be found in Annex G.
Any statistically significant central estimates for differences between patrol days and non-patrol-days (measured by percentage) were converted into estimated volume changes using a counterfactual method. It was assumed with no intervention that the average crime/positive outcome rate on days not patrolled would have applied to all days. The difference between this and the actual number of offences or positive outcomes therefore provides an estimate for total crimes prevented or positive outcomes increased. This was done force by force and then combined into a pooled estimate with a CI using Monte Carlo simulation (see Annex A for more details).
One force was not represented in the results as they were unable to supply data that separated proactive patrolling from reactive entries into the hot spots following crimes or incidents.
It is also standard in studies of hot spot policing to evaluate displacement and diffusion, the degree to which police presence in hot spots changes crime levels in the areas immediately around those hot spots. It is possible that crime might be displaced to surrounding areas, but it is also possible that there will be a ‘diffusion of benefits’ to nearby areas as offenders, seeing additional visible police presence, decide to refrain from offending in an area even larger than the patrolled hot spot. Weisburd and others (2006) found that crime displacement in hot spot policing areas was small and that the diffusion of crime control benefits was more likely. Indeed, the Braga (2019) review of hot spot policing found no consistent evidence for displacement, and that the overall positive diffusion effect was statistically significant but relatively small compared to the effect in the target area.
The displacement/diffusion analysis was carried out by putting a 50m buffer around all hot spots. A buffer is the area immediately surrounding a hot spot up to a given distance. Various buffers are possible, but we choose 50m by considering factors such as proximity of hot spots (often they were only 100m or so apart in city centres), line of sight and density of crimes. The buffer zones were created using R code and constructed so as not to overlap, dividing the potential overlapping space equally, attributing crimes recorded in a buffered area to the nearest hot spot. An example of how an overlapping buffer was converted into a non-overlapping one is shown below:
The analytical approach for the buffer model is the same as used in the main model (see Annex A for full details). The only difference is that instead of comparing crime rates on patrolled days versus non-patrolled days in the hot spot, the buffer model compared these rates within the buffer areas. The same control variables were also used.
Note that the buffered model was only run for the year ending March 2022, because displacement/diffusion was not tested in the evaluation report for that year. We therefore plug that evidence gap here. The model was not run for the year ending March 2023 because there was no significant impact on crime in the hot spots themselves. Table 1.2 shows forces included in the analysis.
Table 1.2: Details of forces included in the displacement/diffusion analysis
Forces included in model | Total hot spot area | Hot spots in model |
---|---|---|
Police Force 1, Design B | 15 | 15 |
Police Force 11 (RCT) | 77 | 28 |
Police Force 14 (RCT) | 98 | 20 |
Police Force 2 (RCT) | 12 | 12 |
Police Force 4 | 21 | 5 |
Police Force 6, Design C | 168 | 12 |
Police Force 7 | 18 | 12 |
There are fewer forces included in the buffer model compared with the main hot spot model. Forces that could only supply data at ward level were excluded because generally those forces were not attempting to patrol the whole ward; this was just the lowest level that the data could be supplied. That means a buffer around the ward would not be a good reflection of the area immediately surrounding the patrolled area.
The analytical approach to cost benefit is set out along with the results in Section 2.3.
HO has also arranged learning events for forces to share how they have addressed challenges more widely and examples of best practice, as well as sharing initial findings at an England and Wales level. These have been instrumental in overcoming challenges, improving the programme during its development, and to make best use of funds and assess impact effectively.
1.5 Limitations to the evaluation findings
HO designed the evaluation to be as robust and comprehensive an approach as possible. However, the approach had some limitations which are important to bear in mind when considering the evaluation findings in this report. Key limitations to the data collection and analysis are listed below.
Treatment and control day definitions
Treatment and control days were defined for each area and day, and recorded crime counts were aggregated to this level. HO designated a day as treated irrespective of the time of day that the patrol occurred. Similarly, a day was defined as a control provided it did not receive a patrol, irrespective of the time that had elapsed since its last patrol. For example, a visit at 23:00 hours would mean that day was considered treated and subject to the same treatment effect for the entire day, including the 23 hours before the patrol began.
Residual deterrence
A fundamental limitation of the repeat crossover experimental design and DODO model was residual deterrence. That is, if treatment on a given day affects crime rates on subsequent unvisited control days, then the treatment effect will be diluted (Sherman, 2022). The wider evidence on residual deterrence is unclear, especially in UK studies (Barnes and others, 2020; Basford and others, 2021; Bland and others, 2021). Residual deterrence has not been assessed in this evaluation, but it is hoped that it will be in future iterations.
Accuracy of crime location
Crime was mapped to a patrolled area using x-y co-ordinates. Uncertainty in the crime location may have resulted in crimes being mis-allocated into or outside a hot spot area. This risk was more acute in smaller areas where a greater proportion of crimes were near the area perimeter, or low crime rate areas that were more likely to be sensitive to specific crime events. In addition, as part of the crime mapping process, adjacent areas which share a boundary may double count crimes. Some crime recording systems used gazetteers that accentuated these issues.
Accuracy of crime date and time
Some recording systems automatically populated date and time of offence if no other information was recorded. This means there were recorded increases of offences on default dates, many of which will not relate to actual numbers of offences. These challenges limited the scope for widening the analysis to consider time-of-day analysis. However, HO accounted for some challenges by adding controls to the model for the first of the month, to limit the impact of the recording bias (see Annex A).
Accuracy of patrol location
HO assumed a patrol would uniformly treat and be limited to the entire designated area. Patrols that were visible outside of the patrol location, including travel to/from the area, could have affected crime rates in the surrounding areas and potentially other nominally untreated hot spots.
Quality of patrol data
Forces used a variety of methods for collecting patrol data. Some forces provided raw GPS data, some forces collated data manually, and some used a mixture of automated and manual (see Table 2.3). The quality and completeness of patrol data; therefore, varied by force. Additionally, the absolute number and proportion of control and treated days varied between hot spots and PFAs, especially when conditioned on co-variates. This may have led to a subset of areas dominating the result and sparse data bias.
Experimental conditions/contamination
The Grip programme ran alongside other normal police activities and other interventions which may have coincided with Grip treatment and control area/times. The crossover design should have mitigated most of these concerns because other interventions were unlikely to only affect certain days within the same area. HO assumed that scheduled visitation was additional to normal levels of police patrols and response.
Pooling results across multiple experiments and PFAs
The pooled result took an average of each force-experiment evaluation. However, there was a large degree of operational freedom in how the police force enacted Grip. Consequently, each estimate was derived from experiments that may have varied significantly in properties, such as the patrol length, number of patrols, patrol area and crime rates.
2. Results
This chapter presents the results, including the delivery metrics, impact evaluation and cost-benefit analysis. Section 1.4 explains the methodological approach used (with Annex A providing additional detail).
2.1 Delivery metrics
Grip forces in the year ending March 2023 carried out at least 90,000 patrols, as shown in Table 2.1. This was 29,059 patrols fewer than the previous year. There are 2 possible reasons for this. Firstly, the overall funding total in the year ending March 23 was £3,305,487 lower than the previous year. Secondly, police forces were encouraged to use funding in the year ending March 2023 for POP alongside their hot spot patrols. As a result, 15% of funding that year was spent on POP.
Table 2.1: Total patrols carried out by funded forces, year ending March 2023
Police force area | Additional patrols from Home Office funding | Funding allocated |
---|---|---|
Avon and Somerset | 4,432 | £717,275.00 |
Bedfordshire | 1,336 | £520,761.00 |
Cleveland | 5,319 | £470,405.00 |
Essex | 6,599 | £713,590.00 |
Greater Manchester | 4,923 | £2,112,522.00 |
Hampshire | 2,244 | £508,479.00 |
Humberside | 7,346 | £439,699.00 |
Kent | 3,605 | £636,213.00 |
Lancashire | 1,598 | £798,337.00 |
Leicestershire | 2,764 | £529,359.00 |
Merseyside | 2,876 | £1,565,969.00 |
Metropolitan Police | 4,351 | £7,988,282.00 |
Northumbria | 4,589 | £967,830.00 |
Nottinghamshire | 2,751 | £591,998.00 |
South Wales | 659 | £480,230.00 |
South Yorkshire | 8,236 | £1,091,879.00 |
Sussex | 3,517 | 488,828.00 |
Thames Valley | 9,151 | £772,545.00 |
West Midlands | 11,313 | £3,023,855.00 |
West Yorkshire Police (WYP) | 4,517 | £1,651,943.00 |
Total | 92,126 | £26,069,999.00 |
Notes:
- The total patrols estimate for Bedfordshire, Essex, Lancashire, Sussex and the Metropolitan Police should be treated as a lower bound, as the structure of GPS returns meant multiple patrols on the same day could not be differentiated.
Force data returns encompassed a wide variety of approaches to conducting and recording hot spot policing. For this reason, direct comparisons should not be made between forces in relation to the levels of activity and funding. For example, forces differed in their patrol lengths, ranging from an average of 15 minutes to 3.5 hours. Different forces also had different numbers of officers within each patrol. The hot spots for the year ending March 2023 also differed in size and the amount of crime contained, as shown in Table 2.2.
Table 2.2. Area and amount of crime covered by the hot spots
Police force area | Hot spots | Area covered by hot spots (km2) | Total force area (km2) | Hot spots as a % of force area | Annual violence/robbery offences in the hot spots | Annual violence/robbery offences in the force | % of offences covered by the hot spots |
---|---|---|---|---|---|---|---|
Avon and Somerset | 34 | 1.18 | 4784.26 | 0.02 | 2543 | 34738 | 7.32 |
Bedfordshire | 10 | 3.22 | 1235.43 | 0.26 | 2275 | 12753 | 17.84 |
Cleveland | 32 | 2.09 | 597.60 | 0.35 | 2283 | 16484 | 13.85 |
Essex | 35 | 1.21 | 3671.51 | 0.03 | 1667 | 41303 | 4.04 |
Greater Manchester | 85 | 26.59 | 1275.98 | 2.08 | 9267 | 86657 | 10.69 |
Hampshire | 25 | 1.56 | 4148.24 | 0.04 | 2812 | 46481 | 6.05 |
Humberside | 98 | 4.49 | 3515.52 | 0.13 | 3018 | 22955 | 13.15 |
Kent | 34 | 50.01 | 3739.23 | 1.34 | 10760 | 50466 | 21.32 |
Lancashire | 30 | 17.41 | 3067.08 | 0.57 | 7990 | 39067 | 20.45 |
Leicestershire | 27 | 1.32 | 2550.88 | 0.05 | 1276 | 23538 | 5.42 |
Merseyside | 13 | 4.22 | 652.19 | 0.65 | 5159 | 43527 | 11.85 |
Metropolitan Police | 75 | 7.79 | 1570.51 | 0.50 | 8290 | 181515 | 4.57 |
Northumbria | 26 | 61.69 | 5572.26 | 1.10 | 8390 | 33171 | 25.29 |
Nottinghamshire | 17 | 38.64 | 2159.32 | 1.79 | 4952 | 23946 | 20.68 |
South Yorkshire | 60 | 1.20 | 1551.55 | 0.08 | 3163 | 34581 | 9.15 |
Sussex | 15 | 0.61 | 3786.65 | 0.02 | 1375 | 30698 | 4.48 |
Thames Valley | 34 | 7.98 | 5743.41 | 0.14 | 6571 | 42155 | 15.59 |
West Midlands | 57 | 12.48 | 901.64 | 1.38 | 10945 | 93682 | 11.68 |
West Yorkshire Police (WYP) | 50 | 3.31 | 2029.26 | 0.16 | 5537 | 69908 | 7.92 |
Overall | 757 | 246.39 | 52552.51 | 0.47 | 98273 | 927625 | 10.59 |
Notes:
- Although the hot spots are those for the programme in the year ending March 2023, the crime totals are for the previous year.
Table 2.2 shows that the forces with the highest volumes of crime are not always the forces with the largest geographical areas. For example, the Metropolitan Police Service (MPS) had the most offences in our violence/robbery grouping in the year ending March 2023 (see Annex G for details of the offences included). But Thames Valley has the largest force area, at 5,743 square kilometres (km2), which is more than 3.5 times the area covered by the MPS. Merseyside has the seventh highest total for crime but the second smallest geographical area. Clearly, the force’s geographical size affects operational efficiency. A force with a larger area will generally spend a higher percentage of time travelling to and from hot spots, meaning a lower proportion of the overall funding is likely to go on patrolling.
Table 2.2 also highlights the degree to which crime is concentrated within hot spots. The programme targets just under 0.5% of the total area covered by our 20 forces, but that accounts for more than 10% of the offences. Arguably, though, the table also reveals an important limitation of hot spot policing. The law of crime concentration suggests that around 50% of crime happens in about 5% of places (Weisburd, 2015). But the programme only covers about 0.5% of total force area, considerably lower. These results imply that a 47% crime reduction in the hot spots would be required to reduce crime by 5% across all the forces. A similar table has been calculated for the year ending March 2022 and is in Annex B.
Forces were encouraged to randomise their patrolling where possible and to track activity in the hot spots via GPS tracking systems, rather than officer-completed forms. Table 2.3 shows which forces used randomisation and GPS tracking.
Table 2.3 Table showing the forces that used randomisation and GPS tracking
Police force area | RCT | GPS |
---|---|---|
Avon and Somerset | Yes | Partial |
Bedfordshire | No | Partial |
Cleveland | Quasi | Yes |
Essex | No | Partial |
Greater Manchester | No | No |
Hampshire | Quasi | No |
Humberside | Yes | Yes |
Kent | No | No |
Lancashire | Yes | Yes |
Leicestershire | Yes | Partial |
Merseyside | Yes* | Yes |
Metropolitan Police | Yes | Partial |
Northumbria | Quasi | Yes |
Nottinghamshire | No | No |
South Wales | No | No |
South Yorkshire | Yes | Yes |
Sussex | Yes* | Yes |
Thames Valley | Yes | Yes |
West Midlands | Yes | Partial |
West Yorkshire Police (WYP) | Yes* | Yes |
Notes:
-
Those labelled ‘quasi’ in the RCT column did a fixed, non-random patrol schedule (see further commentary below).
-
The asterisk in the RCT column indicates that while the bulk of the patrols were generally randomised, a subset of those included in the analysis were not.
-
Those labelled ‘partial’ in the GPS column had some hot spots or some periods of the year that were GPS-tracked.
Table 2.3 shows that 11 forces used randomisation in the year ending March 2023, although in 3 cases, there was also a subset of patrols that were not randomised. Most of those that used randomisation randomised by day, with one force randomising by area. This was an increase compared to the prior year when 9 forces used randomisation. In addition, 4 forces used a quasi-random approach, in which hot spots were patrolled routinely every other day, or similar. This has been labelled quasi-random to distinguish it from forces that developed patrol schedules to target days with the highest level of crime. The latter suffers from a potential endogeneity issue: that treatment days are likely to have systematically higher levels of crime than control days, causing a potential bias against seeing a significant effect from patrols on those days. We employed a basket of control variables to mitigate this.
Table 2.3 also shows that 15 of the 20 forces used GPS tracking to monitor patrols for at least some of their hot spots for at least some of the year. ‘Partial’ cases indicate both instances where a force only had GPS tracking on a subset of hot spots, and instances where they only GPS-tracked patrols for part of the year. This represents a marked step forward from the previous year, when a minority of forces had automated GPS tracking systems in place.
Annex C contains further information on delivery for each of the 20 forces.
2.1.1 Weapons
The aims of the funding were to reduce SV, and therefore while the primary activity for the funded forces was to conduct visible hot spot patrols, forces also conducted additional activity in the hot spots. They often aimed these tactics at increasing visibility, as well as improving safety and reducing SV. This included weapons sweeps, stop-and-search, and holding weapons amnesties. To capture the additional activity, forces were required to submit returns on the number of weapons collected within the force during the funding period. Across the funded forces for the whole financial year ending March 2023, over 85,000 weapons were collected. These were split fairly evenly across the quarters. This is similar to the year ending March 2022, when over 80,000 weapons were collected in funded forces.
Figure 2.1 Total weapons seized during hot spot funding, year ending March 2023 by quarter
Forces varied in their IT systems and recording mechanisms for weapons data. This included collection method (such as seized, surrendered, collected through a weapon amnesty), and type of weapon recorded (knife/sharp instrument, firearm, other). Forces also varied in whether they could identify those solely collected on funded hot spot patrols. As a result, it is not possible to report on collection method, weapon type, and whether the collection can be solely attributed to the funding.
2.2 Impact analysis
The estimated impact of the programme is presented in 3 sections: offences, positive outcomes and positive outcome rates. Each section includes both force-level estimates and pooled estimates for the entire programme. Results discussed below use a statistical significance at the 5% level to reduce the possibility of results occurring by chance.
A definition of crime baskets used, by crime codes and offences, can be found in Annex G. Below, we provide the full results for violence/robbery and possession offences in both table and forest plot formats, additional results for other crime baskets are available in Annex B and Annex F.
Offences:
Unlike in the year ending March 2022, no statistically significant reduction in violence/robbery at the programme level was found, see Table 2.4. The model estimated a confidence interval between -1.6% and 3.5%, with a central estimate of 0.9%. This indicates that, on average and controlling for other factors, violence and robbery were the same on patrol days compared to non-patrol days within the hot spots. Furthermore, at force level, no forces demonstrated a statistically significant reduction in crime. Note that this does not imply crime levels remained the same within the hot spots; the analysis compares patrol days to non-patrol days within the same areas.
Table 2.4: Forest plot and table of violence/robbery offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 11 | -17.99% | 8.72% | -5.58% | 0.425 | |
Police Force 3 | -11.66% | 5.52% | -3.45% | 0.439 | |
Police Force 5 | -21.29% | 18.50% | -3.42% | 0.739 | |
Police Force 9 | -10.23% | 6.15% | -2.38% | 0.573 | |
Police Force 16 | -15.22% | 12.86% | -2.19% | 0.762 | |
Police Force 14 | -10.32% | 12.24% | 0.33% | 0.954 | |
Police Force 1 | -5.20% | 6.25% | 0.36% | 0.901 | |
Police Force 10 | -5.33% | 6.45% | 0.39% | 0.897 | |
Police Force 12 | -10.99% | 13.58% | 0.55% | 0.930 | |
Police Force 15 | -9.81% | 12.88% | 0.90% | 0.876 | |
Police Force 20 | -4.32% | 7.68% | 1.50% | 0.621 | |
Police Force 19 | -4.58% | 9.25% | 2.10% | 0.547 | |
Police Force 18 | -5.87% | 11.36% | 2.38% | 0.583 | |
Police Force 2 | -5.72% | 11.35% | 2.46% | 0.568 | |
Police Force 4 | -4.69% | 12.90% | 3.73% | 0.396 | |
Police Force 7 | -5.09% | 16.07% | 4.96% | 0.346 | |
Police Force 17 | -8.03% | 20.34% | 5.20% | 0.460 | |
Police Force 13 | -4.61% | 17.45% | 5.85% | 0.284 | |
Pooled Average | -1.60% | 3.48% | 0.88% | 0.479 |
Notes:
-
Police Forces 4, 10, 16, 2, 7, 12, 9, 20, 19 and 14 used randomisation in all or part of their patrol schedules.
-
Columns labelled ‘Sig’ throughout these tables denote whether results are statistically significant, with * indicating significance at the 10% level and ** indicating significance at the 5% level.
There was also no significant effect for Police Force 6, which used control areas rather than control days. The effect size was 7.15% (CI: -3.07% to 17.36%) with p-value of 0.17.
Similar results were observed for the other baskets of crime tested, with no statistically significant differences in the number of offences on patrol days. However, the model did find a statistically significant increase in possession offences, with a 19.2% increase on patrol days in the funded forces, see Table 2.5. This means that, on average and controlling for other factors, possession offences were 19.2% higher on patrol days than on non-patrol days in the hot spots. Similar trends were discovered for the year ending March 2022, which is in Annex B.
Table 2.5: Forest plot and table of possession offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 17 | -21.46% | 8.15% | -7.84% | 0.317 | |
Police Force 4 | -20.03% | 31.50% | 2.54% | 0.843 | |
Police Force 12 | -25.43% | 47.90% | 5.02% | 0.779 | |
Police Force 2 | -11.11% | 31.30% | 8.03% | 0.437 | |
Police Force 18 | -10.35% | 32.77% | 9.10% | 0.385 | |
Police Force 9 | -9.18% | 32.75% | 9.80% | 0.334 | |
Police Force 14 | -18.17% | 47.38% | 9.82% | 0.533 | |
Police Force 20 | -0.78% | 29.42% | 13.32% | 0.065 | * |
Police Force 1 | 3.38% | 29.92% | 15.90% | 0.011 | ** |
Police Force 15 | -0.77% | 37.13% | 16.65% | 0.062 | * |
Police Force 13 | -6.88% | 51.28% | 18.69% | 0.166 | |
Police Force 7 | -4.25% | 61.52% | 24.36% | 0.102 | |
Police Force 11 | -17.22% | 92.31% | 26.17% | 0.280 | |
Police Force 10 | 6.03% | 53.91% | 27.75% | 0.010 | ** |
Police Force 19 | 8.81% | 55.66% | 30.14% | 0.004 | ** |
Police Force 5 | -29.72% | 156.18% | 34.18% | 0.373 | |
Police Force 3 | 0.23% | 82.42% | 35.22% | 0.048 | ** |
Police Force 16 | 11.81% | 91.66% | 46.39% | 0.006 | ** |
Pooled Average | 11.49% | 28.60% | 19.15% | 0.000 | ** |
Notes:
- Police Forces 4, 10, 16, 2, 7, 12, 9, 20, 19 and 14 used randomisation in all or part of their patrol schedules.
Consistent with the above, Police Force 6, which randomised hot spots by area rather than by day, had a 22.90% effect for possession (CI: -5.15% to 50.96%) with p-value 0.11. Possession offences are generally taken to be more of a measure of police activity rather than the underlying level of weapons or drugs possession. In that light, it is not perhaps surprising that offences increased in the hot spots when patrolled.
Positive outcomes:
As well as testing whether hot spot policing reduced the number of crimes in the hot spots, we also tested whether it affected the outcomes of offences occurring in the hot spots. A positive outcomes means – in effect – that the crime was solved or cleared. Annex G has a full set of the outcomes included. Table 2.6 shows results for violence/robbery positive outcomes. The pooled estimate was an 8.1% increase (CI: 0.7% to 17.2%) on patrol days compared to non-patrol days in the hot spots, so a greater number of violence/robbery cases are being cleared in the hot spots if they happen on patrol days.
Similar impacts were observed for positive outcomes in other crime categories, including all crimes minus possession and victim-based crimes, with significant increases of 7.8% (CI: 3.7% to 12.4%) and 8.2% (CI: 3.2% to 14.3%), respectively on patrol days. Details of these results are available in Annex F. Positive outcomes for possession offences showed the largest significant increase: 20.0% (CI: 11.5% to 30.8%). Given that this offence grouping comprises weapons possession and drugs possession, this means that the programme’s patrols are leading to more people being caught with weapons and drugs in the hot spots.
Table 2.6: Forest plot and table of violence/robbery offences with a positive outcome, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 16 | -34.02% | 16.39% | -12.37% | 0.362 | |
Police Force 11 | -28.03% | 21.28% | -6.57% | 0.609 | |
Police Force 14 | -29.18% | 27.07% | -5.14% | 0.724 | |
Police Force 20 | -28.67% | 27.72% | -4.55% | 0.754 | |
Police Force 10 | -16.88% | 17.29% | -1.26% | 0.885 | |
Police Force 1 | -16.09% | 18.17% | -0.42% | 0.962 | |
Police Force 12 | -30.80% | 53.67% | 3.12% | 0.880 | |
Police Force 9 | -14.13% | 28.52% | 5.05% | 0.632 | |
Police Force 2 | -21.07% | 41.33% | 5.62% | 0.713 | |
Police Force 15 | -10.93% | 32.84% | 8.78% | 0.409 | |
Police Force 3 | -14.47% | 40.42% | 9.59% | 0.469 | |
Police Force 13 | -20.95% | 54.01% | 10.34% | 0.563 | |
Police Force 4 | -13.49% | 41.42% | 10.61% | 0.421 | |
Police Force 17 | -10.81% | 46.49% | 14.31% | 0.291 | |
Police Force 7 | -10.53% | 48.14% | 15.13% | 0.274 | |
Police Force 19 | -6.34% | 47.41% | 17.50% | 0.163 | |
Police Force 18 | -6.69% | 49.27% | 18.01% | 0.167 | |
Police Force 5 | -25.28% | 159.51% | 39.25% | 0.297 | |
Pooled Average | 0.68% | 17.24% | 8.10% | 0.032 | ** |
Consistent with the above, Police Force 6, which randomised hot spots by area rather than by day, had a 39.29% effect for possession (CI: -0.91% to 79.49%) with p-value 0.055.
Table 2.7: Forest plot and table of possession offences with a positive outcome, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 12 | -30.88% | 35.10% | -3.36% | 0.841 | |
Police Force 17 | -14.66% | 18.32% | 0.48% | 0.954 | |
Police Force 20 | -17.20% | 30.30% | 3.87% | 0.743 | |
Police Force 4 | -26.70% | 50.95% | 5.19% | 0.784 | |
Police Force 1 | -7.88% | 24.73% | 7.19% | 0.369 | |
Police Force 2 | -3.95% | 24.45% | 9.33% | 0.177 | |
Police Force 13 | -22.78% | 64.91% | 12.85% | 0.532 | |
Police Force 15 | -5.87% | 36.16% | 13.21% | 0.188 | |
Police Force 9 | -3.24% | 36.91% | 15.10% | 0.112 | |
Police Force 18 | -6.40% | 42.28% | 15.40% | 0.180 | |
Police Force 7 | -17.68% | 63.09% | 15.87% | 0.398 | |
Police Force 10 | -8.37% | 46.76% | 15.97% | 0.218 | |
Police Force 14 | -13.38% | 76.91% | 23.79% | 0.241 | |
Police Force 5 | -32.72% | 178.23% | 36.82% | 0.387 | |
Police Force 19 | 16.63% | 64.17% | 38.37% | 0.000 | ** |
Police Force 16 | 3.26% | 88.10% | 39.37% | 0.030 | ** |
Police Force 11 | -1.23% | 102.35% | 41.38% | 0.058 | * |
Police Force 3 | 5.32% | 98.99% | 44.77% | 0.023 | ** |
Pooled Average | 11.45% | 30.79% | 20.01% | 0.000 | ** |
Again, the result for Police Force 6 was consistent with the pattern in other forces. Using an area-based comparison, it had an effect size of 17.21 (CI -11.62 to 46.05) with p-value 0.24.
Positive outcome rates:
Tables 2.8 and 2.9 show the results for positive outcome rates for violence/robbery and ASB offences. For violence/robbery, there was a non-statistically significant increase of 7.2% (CI: 1.1% to 18.2%) in the hot spots on patrols days. The result for ASB offences was similar: a non-statistically significant increase of 5.4% (CI: -2.8% to 15.4%). There was a significant increase in the rate of positive outcomes for victim-based crimes and all crimes minus possession: 8.8% (CI: 3.2% to 15.9%) and 8.3% (CI: 3.5% to 14.0%) respectively. This means the rate of solving victim-based and all crimes minus possession statistically increased on patrol days within hot spots.
Table 2.8: Forest plot and table of positive outcome rate for violence/robbery offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 12 | -45.65% | 21.71% | -18.67% | 0.315 | |
Police Force 20 | -35.94% | 22.92% | -11.26% | 0.472 | |
Police Force 11 | -31.51% | 21.84% | -8.65% | 0.538 | |
Police Force 10 | -20.96% | 11.52% | -6.11% | 0.473 | |
Police Force 1 | -22.22% | 21.34% | -2.85% | 0.799 | |
Police Force 7 | -28.30% | 33.43% | -2.19% | 0.889 | |
Police Force 14 | -22.22% | 37.92% | 3.57% | 0.810 | |
Police Force 9 | -16.09% | 28.24% | 3.73% | 0.735 | |
Police Force 2 | -21.66% | 43.87% | 6.16% | 0.700 | |
Police Force 17 | -11.50% | 28.34% | 6.58% | 0.502 | |
Police Force 4 | -19.49% | 42.34% | 7.05% | 0.639 | |
Police Force 13 | -22.11% | 48.95% | 7.71% | 0.653 | |
Police Force 16 | -23.41% | 51.88% | 7.85% | 0.665 | |
Police Force 15 | -7.41% | 37.23% | 12.72% | 0.233 | |
Police Force 3 | -16.82% | 67.52% | 18.04% | 0.353 | |
Police Force 18 | -6.40% | 50.00% | 18.49% | 0.159 | |
Police Force 19 | -5.88% | 63.74% | 24.14% | 0.126 | |
Police Force 5 | -36.14% | 200.53% | 38.53% | 0.409 | |
Pooled Average | -1.08% | 18.16% | 7.19% | 0.090 | ** |
Consistent with the other forces, Police Force 6 had an effect size of 28.67% (CI: -8.97% to 66.31%) with p-value 0.14, using an area-based comparison.
Table 2.9: Forest plot and table of positive outcome rate for ASB offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 2 | -33.19% | -5.58% | -20.58% | 0.009 | ** |
Police Force 13 | -41.90% | 8.74% | -20.51% | 0.151 | |
Police Force 3 | -40.12% | 15.04% | -17.00% | 0.263 | |
Police Force 14 | -33.10% | 13.65% | -12.80% | 0.311 | |
Police Force 5 | -60.21% | 105.54% | -9.56% | 0.810 | |
Police Force 12 | -31.41% | 29.41% | -5.78% | 0.713 | |
Police Force 4 | -25.08% | 28.11% | -2.03% | 0.881 | |
Police Force 16 | -28.75% | 36.49% | -1.38% | 0.933 | |
Police Force 18 | -10.07% | 15.00% | 1.69% | 0.789 | |
Police Force 1 | -8.46% | 14.89% | 2.55% | 0.663 | |
Pooled Average | -2.75% | 15.39% | 5.41% | 0.200 | |
Police Force 19 | -21.13% | 43.07% | 6.23% | 0.691 | |
Police Force 17 | -30.13% | 66.08% | 7.72% | 0.736 | |
Police Force 10 | -14.70% | 40.63% | 9.52% | 0.476 | |
Police Force 20 | -15.60% | 49.96% | 12.50% | 0.422 | |
Police Force 15 | -9.55% | 42.29% | 13.45% | 0.275 | |
Police Force 9 | -2.20% | 47.07% | 19.93% | 0.081 | * |
Police Force 7 | -3.45% | 91.34% | 35.92% | 0.079 | * |
Police Force 11 | -4.79% | 143.84% | 52.37% | 0.079 | * |
The result for Police Force 6, using an area-based comparison, was an effect size of -4.25% (CI: 31.49% to 22.99%) with p-value 0.76.
Year ending March 2022 expanded results
Results for the year ending 2022 have also been updated and expanded to include the estimated impact of patrols on positive outcomes and positive outcome rates in hot spots. Full results can be found in Annex B. This brief section provides a summarised version and compares the results with the year ending March 2023.
Offences:
A revised result was generated for the main outcome of interest, violence/robbery offences. This showed a statistically significant -5.2% reduction on patrol days compared to non-patrol days in hot spots (CI: -8.9% to -1.3%), slightly lower than the 7% estimated in the previous publication. There are a few reasons for this difference, including the removal of filters for low-crime, low-patrol hot spots and revised crime data (see Annex A). Other than that, the results were similar to those for the year ending March 2023, with no significant effects for other crime baskets except possession, which showed a 24.3% increase on patrol days.
Positive outcomes:
Similar to the year ending March 2023 results, a statistically significant increase in positive outcomes was found for possession offences, with the model estimating a 50.2% increase (CI: 23.5% to 117.0%) on patrol days compared with non-patrol days in the hot spots. Unlike the results for the year ending March 2023, there were no other statistically significant increases in positive outcomes for the other crime baskets. However, all effects were in the direction of an increase.
Positive outcome rate:
There was non-statistically significant increase in the rate of positive outcomes for violence/robbery offences on patrol days, with the model estimating a 26.1% rise (CI: -1.9% to 239.1%). The large CI is due to one force with a lower number of positive outcomes within their hot spots. This result was; however, significant at a 10% confidence level, which could suggest that the absence of a significant increase in positive outcomes for these offences above was likely due to the crime reduction effects. For ASB offences, there was a non-statistically significant increase in the rate of positive outcomes of 15.6% (CI: -9.2% to 128.8%).
Taken together, the results across the 2 years show a reasonably consistent impact on positive outcomes: the programme is leading to more offenders being caught and more weapons and drugs being found in the hot spots on patrol days. Possible reasons for these results and some of their implications are analysed in Chapter 3.
2.3 Innovation Fund impact analysis
Five forces received money from the Innovation Fund. Police Force 6 trialled fixed-post policing at hot street junctions and the use of drones in a highly visible manner in serious violence/robbery hot spots. This activity was evaluated externally. Police Force 10 deployed plain-clothes observers to target potential offenders and deployed volunteer guardians to provide support and advice in the night-time economy. This activity could not be evaluated because of insufficient data. Police Force 12 employed additional door supervisors to supplement patrols in the night-time economy. As Police Force 12’s activity was similar to core Grip hot spot policing but delivered by partner agencies, a combined impact result for Innovation Fund and core Grip activity is included in Section 2.2. Police Force 16 deployed street outreach activity in hot spots, focusing on diverting young people into services. The Home Office evaluated this activity. Police Force 19 deployed uniformed and plain-clothes officers alongside a targeted social media campaign focused on crime prevention. The Home Office evaluated this activity.
Table 2.10: Innovation Fund results
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 19 | -16.7% | 0.0% | -8.7% | 0.05 | ** |
Police Force 16 | -20.7% | 22.9% | -1.3% | 0.91 | |
Police Force 6 (external) | -13.3% | 4.9% | 23.1% | 0.61 |
A statistically significant reduction in violence/robbery of -8.7% was found for Police Force 19’s Innovation Fund activity. This means that crime was 8.7% lower in hot spots receiving the intervention than in comparable control areas not receiving the intervention (during the intervention period, relative to the period before). However, it should be noted that this intervention has a low sample size. This force’s Innovation Fund activity was a combination of uniformed and plain-clothes patrols to spot subjects acting suspiciously within the hot spots, and a targeted social media campaign.
There was no significant effect on crime on treatment days compared to control days in Police Force 16, where street outreach workers were deployed. It should be noted that other hot spot activity may have occurred on Innovation Fund control days.
Police Force 6’s fixed-post policing and drones intervention was evaluated externally. The evaluation produced 108 results across different variables, including crime type (any offending, public space offending, public space violence), measure of crime (crime days, crime volume, crime harm), geographical unit (hot spot, hot spot with buffer, fixed point of the intervention with buffer) and analysis time period (hours of intervention only, full day). The result presented in Table 2.10 is the result that most closely matches the other evaluation results presented in Section 2.2 and [Section 2.3.}(#s2-3). This shows there was no significant impact on public place violence volume in the hot spot compared to control areas.
The only significant result found in this evaluation was a 74.5% significant increase in public space violence within 25m of the fixed point of intervention. This may be an increased reporting of crime in the area immediately around a fixed-post police officer; however, the small geographical area means sample sizes for this result are very low and should be treated with caution.
2.4 Displacement/diffusion of benefits results
The displacement/diffusion analysis was carried out on a subset of forces using data from the year ending March 2022, for reasons outlined in the methodology section. Table 2.11 shows that there was a significant impact on violence/robbery in those hot spots. In fact, the effect is greater – an 11% reduction in violence/robbery on days patrolled versus days not within the hot spots – than was found for the full set of hot spots (-7%).
Table 2.11: Impact on violence/robbery within the hot spots, for the subset of forces used in the diffusion/displacement analysis
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 1, Design B | -19.6% | 5.4% | -8.0% | 0.23 | |
Police Force 11 (RCT) | -13.8% | 8.1% | -3.5% | 0.54 | |
Police Force 14 (RCT) | -22.6% | 22.5% | -2.6% | 0.82 | |
Police Force 2 (RCT) | -38.5% | 7.7% | -18.6% | 0.15 | |
Police Force 4 | -48.1% | -8.0% | -30.9% | 0.01 | ** |
Police Force 6, Design C | -12.8% | 3.9% | -4.82% | 0.27 | |
Police Force 7 | -21.1% | 5.5% | -8.8% | 0.21 | |
Pooled average | -18.5% | -3.6% | -11.0% | 0.00 | ** |
Table 2.12 shows the impact in the 50m buffer zones for the same forces and hot spots.
Table 2.12: Impact on violence/robbery within the buffer zones for the subset of year ending March 2022 hot spots suitable for diffusion/displacement analysis
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 1, Design B | -9.8% | 42.4% | 13.3% | 0.28 | |
Police Force 11 (RCT*) | -26.5% | 16.7% | -7.4% | 0.52 | |
Police Force 14 (RCT) | -11.1% | 65.1% | 21.1% | 0.23 | |
Police Force 2 (RCT) | -40.3% | 12.2% | -18.2% | 0.21 | |
Police Force 4 | -48.6% | 16.2% | -22.7% | 0.22 | |
Police Force 6, Design C | -5.0% | 11.3% | 2.81% | 0.49 | |
Police Force 7 | -32.7% | 82.6% | 10.9% | 0.69 | |
Pooled average | -12.2% | 12.1% | 0.0% | 1.00 |
Notes:
-
RCT* denotes cases where randomisation was involved but only for part of the period, or some of the areas or had a known issue, such as where an element of officer choice compromised the randomisation process to some degree. Note that the other RCT experiments also had varying levels of compliance, which can lead to randomisation issues, but this was mitigated by still using day and month controls within our models.
-
Statistical significance is denoted by the ‘Sig’ column with * indicating significance at the 10% level, ** at the 5% level and *** at the 1% level.
None of the effects are statistically significant and the pooled average is exactly zero, suggesting no displacement or diffusion of benefits. In other words, the significant violence/robbery reduction achieved in the hot spots was not achieved by simply pushing the crime into adjacent areas that did not receive patrolling. Annex B shows the sample sizes for these models.
2.5 Cost-benefit analysis
This section aims to compare the costs of the programme in the year ending March 2023 against its benefits. As the results from our meta-analyses did not show a statistically significant change in crime incidents for the year ending March 2023, these have not been monetised for inclusion in the cost-benefit analysis. Significant impacts were observed for positive outcomes (a measure of crime clearance). However, at this point, the Home Office does not have an established method for quantifying these impacts. As a result, this section focuses mainly on the costs of funding the hot spot activity evaluated in this report, although we briefly discuss the implications of the non-monetised impacts at the end.
The estimated costs of the programme can be broken down into 3 main components:
- funding allocations to the police forces
- HO and other non-police staff involved with delivering or evaluating the programme
- College of Policing funding to deliver support to forces
Section 2.1 shows the force allocations for the year ending March 2023, totalling £26.1 million. For the cost-benefit analysis, not all the allocated spend was included. This was because some of the funded activity could not be evaluated and hence could not generate a benefit. Therefore, HO excluded these cases, which total £9.8 million. Examples include:
- several forces began POP activity in the year ending March 2023, separate to the main visible patrols, and POP has not been evaluated at this stage
- some forces funded more than one hot spot operation with their allocation but only provided data for evaluation on one of the operations
- due to data issues in tracking officer patrols, one force could not be included in the analysis
As forces run programmes over multiple years, the period we are evaluating does not always line up perfectly with financial years. To account for this, we have adjusted the financial spend to include the amount spent over the whole period. For example, if a force changed their hot spots in January to March 2022 and continued with these into the following year then we have included these in the year ending March 2023 cost. In this case, we have adjusted the allocation for the force to include the additional spend. We did this by looking at the proportion of patrol days in the ‘spillover’ period relative to total annual patrol days. Based on this ratio, costs were re-apportioned across years.
An additional £1.5 million was allocated to forces through the Innovation Fund.
HO staff costs were estimated using internal wage and time-spent figures. The London School of Economics and Political Science was also involved, providing guidance on the evaluation method, so the cost of their time was also included. Taken together, this added a further £0.6 million in costs.
Excluding the ‘non-evaluated’ spends and adjusting for spend in different years, the remaining allocation cost of the programme was £16.3 million. When adjusted to financial year 2025 prices using the HMT deflator, the total cost was £17.5 million.
As discussed above, it was not possible to monetise any benefits for the programme in the year ending 2023, meaning that unlike for the previous year, we cannot say that the programme was cost effective in the year ending 2023. However, it is worth noting that the total crime reduction benefits in the previous year were estimated at £38.4 million with programme costs totalling £16.8 million, in financial year 2025 prices. That means that overall, the programme’s crime reduction benefits (£38.4 million) still exceed the total cost across both years (£34.3 million), even without monetising the impact on positive outcomes, discussed briefly below.
Unmonetised benefits
The results for the year ending 2023 show that positive outcomes saw a statistically significant increase of 8.7% for SV, an increase of 7.9% for all crimes (minus possession) and an increase of 20.7% for possession offences (comprising weapons and drugs possession) in the hot spots on days patrolled. In other words, the programme seems to be causing more crimes to be solved and more weapons/drugs to be found.
It is hoped that in the future we will be able to monetise the costs and benefits of these impacts. Though it is likely these would be net benefits for the programme, it is worth noting that these impacts incorporate both costs and benefits. On the benefit side, higher clearance rates might act as a deterrent for other offenders and as a direct deterrence for those receiving the sanctions. Benefits might also arise from fine income and, if the additional outcome is a custodial sentence, there might also be an incapacitation effect in which an offender cannot offend while in prison. It is also possible that there may be post-prison benefits of rehabilitation and employment. However, increased crime clearance also creates costs to the police and criminal justice system, notably any court time or prison requirements.
Separately, it should also be noted that we have not been able to test for other effects such as community impacts, for example perceptions of crime and the police.
3. Discussion
This evaluation is based on the second year of centrally funded hot spot operations for SV across England and Wales. The impacts on crime visible in the pilots and year one of the programme have not been carried through into year 2. At the same time analysis has shown that there has been a consistent effect on positive outcomes (crimes cleared) in the hot spots across both years of the programme. This section examines possible reasons why the impact on crime may have waned and why the programme may be generating a more consistent impact on crimes cleared. Although no formal process evaluation was conducted in the year ending March 2023, this section does contain insights gained from our regular discussions with the participating forces.
3.1 Why the impact on crime may have waned
Implementation may have reduced or worsened over time
As Table 2.1 showed, there were fewer total patrols delivered by the programme in the year ending March 2023 compared with the year prior. Partly this was due to having slightly lower levels of funding and partly it was due to forces spending more of the resource on problem-solving activity rather than standard visible patrols. We also heard from some forces that there was a degree of operational fatigue, particularly for those forces using overtime to staff the patrols. Several forces noted that the overtime slots were very popular at first but became harder to fill as time went on. Several forces also noted that it became harder to resist ‘abstractions’, which are cases in which the hot spot patroller was called away due to urgent business elsewhere.
One force also told us that it had become hard to retain the focus on the programme, because of new initiatives arising, both from the Home Office and from within the force itself. On the other hand, although patrol numbers decreased, the programme still delivered more than 90,000 additional patrols that – based on outside evidence – we would still expect to have a significant impact. Furthermore, an increasing number of forces used GPS systems to track some or all of those patrols, meaning we can have high confidence that they were delivered as suggested.
Hot spot policing may become less effective over time even if implementation is maintained
Most of the existing evidence on hot spot policing comes from studies lasting a year or less. We do not therefore really know whether the impacts of hot spot policing can be maintained over the long-term even if the patrols continued to be implemented correctly. There are a number of theoretical reasons why results could drop off over time even if delivery is maintained. For example, it is possible that offenders may adapt. Once they get used to the higher levels of police presence, and possibly even get to know the pattern and timing of (non-randomised) patrols, they may find new ways to offend at their old level. Another possibility is that crime gets lowered to a level where it becomes harder to see new statistically significant impacts. Perhaps the success in year one made success in year 2 impossible? These are possible.
However, the consistent increase in crimes being cleared in the hot spots perhaps argues against the first explanation. More offenders are being caught and punished – wouldn’t we expect this to continue to reduce crime even in the presence of adaption? It’s also worth pointing out that many forces refreshed or completely changed their hot spots from the first year of the programme to the second.
Increased problem-solving activity, or other interventions, may have diverted resources or attention from the patrols
As noted above, in the first year of the programme, forces concentrated predominantly on setting up their regular patrols. In the year ending March 2023, many attempted to add a problem-solving element. General feedback from forces was that this was considerably more difficult and time-consuming to implement than the standard patrols, which is in line with wider academic evidence (see for example, Braga and Weisburd, 2006). Trained problem-solving analysts were not always available and making a good link between the analysis of the problem and a suitable solution was often difficult. Attitudes towards problem-solving varied considerably within the officers tasked to do it. As such, it is possible that the attention required to launch problem-solving distracted from the management of the patrol programme in such a way as to reduce its effectiveness.
Whether this was still a price worth paying remains to be seen. Further work to evaluation of the funded problem-solving activity would resolve that. For now though, it’s worth pointing out that results were not systematically better in forces that did not start problem solving in the year ending March 2023. It’s also worth pointing out that an alternative version of this theory would be that other interventions aimed at reducing serious violence were also introduced or matured in the year ending March 2023. For example, Clear, Hold, Build; Violence Reduction Units; Focused Deterrence pilots; Serious Violence Reduction Orders; County Lines activity and so on. These would, we would hope, decrease violence/robbery generally, including in the hot spots, perhaps making it harder to detect a difference on patrol days versus days not patrolled.
Our methodological approach may be less effective at detecting impact over time
The crime-impact analysis in this report uses a model that compares crimes on days patrolled against crime on days not patrolled. It’s possible that the programme could still be causing crime reduction in the hot spots, even if the difference between patrol days and non-patrol days is not significantly different. One reason is that the deterrent effect of patrols may last from one day into another. This effect, so called ‘residual deterrence’ is perhaps most obvious for a patrol that happens at, say 11pm on a Friday night, where it seems distinctly possible that any deterrent effect would last through to the early hours of Saturday morning. But it is also possible that residual deterrence lengthens and generalises as the programme continues. That is, that offenders get used to a higher level of police presence generally and reduce offending at all times, not just when they see officers on patrol, or shortly after that.
To test whether this is the correct explanation we are building a model to detect residual deterrence, and an alternative evaluation model that tests impact against matched control areas. These should show whether the reason for the lack of impact is methodological. For now, we note only that both the initial pilot studies and at least one influential international study used the crossover design employed here and revealed large impacts from hot spot policing. So, we know that this method can and does pick up effects. The question is whether the mechanism of the effect changes over time.
As residents of hot spots get used to seeing more officers, they may feel confident reporting more offences
This explanation suggests that crime was still reduced but was balanced by an increase in reporting of crime. There are a few nuggets of information that support this hypothesis. A 2024 US study has shown that reporting effects may cause preventative policing interventions to be under-estimated, (Weisburd et al., 2024). Furthermore, some forces did report on offences that were ‘police-reported’ within the hot spots, and some did find that police-reported offences increased on the days police were there relative to the days they were not. In one example, in which a force employed fixed-post (that is, unmoving) officers at key points within the hot spot, they found a mixed effect on the hot spot as a whole but a significant increase in offending within the 25 metres immediately around the officer.
It seems hard to explain this in any way other than a reporting/recording effect. But whilst this explanation seems logical, it is odd that reporting/recording effects do not seem to have affected most other hot spot studies to any great degree. As such, it feels like this explanation only works if it can be shown that people become more likely to report offences the longer police continue to be visible and present in their area. This is therefore a potentially fruitful line of further research.
Our judgment is that waning and uneven implementation is the most likely of these explanations, although reporting effects may be important too. Some forces were quite candid about the difficulty of maintaining focus on the hot spot programme over time, including supervision. Partly this was due to ‘overtime fatigue’ as fewer officers signed up for the extra hot spot shifts; but partly it was due to new initiatives taking the focus away from hot spot policing. To try and tackle this, HO conducted a series of meetings with forces at the end of the year ending March 2023 in order to feed back performance data in the hope of improving implementation in the following year. This involved analysing coverage across hot spots and pointing out those areas that had received fewer patrols. It also involved analysing whether hot spots were being attended at the right time of day.
To investigate possible reporting effects, we attempted to more systematically remove crimes reported by the police directly, on the basis that these could be biased upwards by hot spot patrols. However, only a minority of forces return this data to HO and the data quality was variable. As such, the results of this analysis were inconclusive.
3.2 Why hot spot policing may be increasing positive outcomes (crimes cleared or solved)
The impact of hotspot policing on positive outcomes has been less thoroughly investigated than its impact on crime reduction, but there is evidence that hot spot policing can impact arrests. For example, the Braga et al (2012) meta-analysis noted an increase in ‘opportunistic arrests and investigations’ in the following hotspot experiments:
-
an assessment of Operation Sunrise, which targeted drug corners in Philadelphia, burdened the local judicial system by producing a high number of arrests, (Goldkamp and Vilcica, 2008)
-
similarly, the Vera Institute of Justice’s assessment of the Tactical Narcotics Teams found that the treated 70th Precinct experienced a higher arrest rate compared to the 67th Precinct, which had reallocated some uniformed officers to a different initiative, (Sviridoff et al., 1992)
-
moreover, the Kansas City Gun Project, which studied the impact of proactive patrols and rigorous enforcement of firearm laws, reported an increase in arrests compared to routine police activity, (Sherman and Rogan, 1995)
This section examines possible mechanisms for the impact, ordered roughly by how likely we think they are.
For possession offences, it is likely that officers are searching more people in hotspots and finding more weapons and drugs. Results revealed a very consistent increase in positive outcomes for possession across forces and across both years of the programme. One possible explanation is that stop-and-search tactics are being routinely employed in hotspots during the patrols and that more weapons and drugs are being found as a result. This could be tested by comparing stop and search rates in the hot spots on days patrolled versus days not. We hope to include this analysis in a future iteration of this report.
There is evidence that response times affect outcomes, so by being on the scene more often, hot spot policing may reduce average response times. With quicker response times, officers may prevent more suspects from escaping, gathering more witness statements, and securing crime scenes faster. All of which are likely to increase the solving of crimes. Research conducted by Kirchmaier and Vidal (2018) strongly suggests that faster response time was an effective means of apprehending a larger percentage of offenders. The key mechanism noted for this, was that a faster response time increased the likelihood that a suspect will be named by a victim or witness.
Similarly, it is possible that officers witness more offences directly by being on scene, making offences easier to solve. Police presence could mean witnessing crime firsthand which would greatly increase the chance of apprehending suspects. This would also lead to direct evidence and eyewitness accounts from the officers.
Officers may pick up more intelligence via on-the-ground patrols that help them to solve more offences. The presence of officers during a patrol may help gather evidence, both in what they observe and through engagement with members of the public. Bittner (1970) also suggests that officer’s familiarity with their patrol areas helps them develop an intuitive understanding of normality in their patrol area. This allows them to recognise non-normal behaviour and intervene. Through gathering valuable intelligence and becoming more familiar with local environments, officers may therefore solve more crimes. Mostly, this would drive improved positive outcome rates on all days, but results for patrol days could be even better if the gathered intelligence relates to crimes that have just happened.
Hot spot policing may improve police/community relations, increasing the likelihood that witnesses will come forward. Officers may be creating stronger relationships with their community; for example, some forces actively engage businesses and community groups. Positive interactions may also foster trust and cooperation, making the community more likely to share information and report crime, and especially so when officers are on patrol. Again, this explanation is less likely to create a statistical difference on positive outcomes between patrol and non-patrol days, and more likely to generate improved outcomes on all days (which might be shown via an area-based comparison).
It is possible that crimes reported on patrol days are being processed differently and subsequently solved more effectively. For instance, crimes on patrol days may be more likely to receive an in-person officer response and interview. This could occur because officers are already in the area or because the public directly reports the crime to them. As a result, these cases might be more thoroughly investigated and resolved. The reasoning of this is difficult to determine, it could be attributable to the emotional impact on officers who witness the victims and the consequences of the offense first hand or officers may be inclined to investigate and ask questions when already at the scene.
4. Conclusion
4.1 Summary of findings
This report includes impact evaluation results for the second year of the Grip and bespoke-funded hot spot policing programme (year ending 31 March 2023), including Innovation Fund projects. It also revises and expands on the results for the previous year.
The programme delivered at least 90,000 patrols in the year ending March 2023, meaning that around 220,000 additional patrols were delivered across the first 2 years of the programme. The number of additional patrols decreased slightly from the first year to the second, partly due to a slight drop in funding and partly due to greater spending on problem-solving activity in year 2.
The programme recorded more than 85,000 weapons collected in the year ending March 2023, about 5,000 more than were recorded collected in the year ending March 2022. There were also increases in the number of forces using randomisation to gauge impact of the patrols (from 9 in year 1 to 11 in year 2) and in the number of forces using GPS technology to wholly or partially track patrols, which is now up to 15 out of the 20 forces.
There were 757 hot spots targeted across the 20 forces, accounting for just under 0.5% of the total force area but just over 10% of our basket of violence/robbery offences (details of the exact offences included are in Annex G).
Results for the meta-analysis did not suggest that the programme had a statistically significant impact on volumes of violence/robbery offences in the hot spots in the year ending March 2023. This differs from the 7% reduction reported for the programme the previous year. However, there was a statistically significant violence/robbery reduction for one of the Innovation Fund projects.
There may be several reasons for the lack of programme-level impact. It is possible that hot spot policing was not implemented as well in the year ending March 2023 as in previous years. It is also possible that hot spot policing effects wane over time. While the existing evidence base for hot spot policing is very strong, most studies do not test the effects for longer than a year. Another possibility is that patrolling became less effective because forces put more of their efforts and spent more of their funding on POP in the year ending March 2023. It is also possible that our methodology may become less effective at detecting an effect over time. Our model compares days patrolled versus days not patrolled. Over time, it is possible the accumulation of patrols causes a general deterrence effect that lowers crime on all days, regardless of whether there is a patrol.
This report also estimated the effect for other types of crime. Generally, we did not find significant impacts in either year for other types of crime. In one sense, this is not surprising. The hot spots were selected because of their high levels of serious violence, not because of their levels of other crimes. However, it is worth noting that many other hot spot patrol studies have found effects on a wide variety of crime types.
Results did not suggest any displacement of crime to nearby areas, nor any diffusion of benefits. This was tested using a 50m buffer area for a subset of the year ending March 2022 results, in which we found an 11% reduction in violence/robbery in the hot spots themselves. It was not tested for the following year because there was no significant effect in the hot spots. The lack of evidence for displacement is consistent with other findings in the literature.
While this study did not find an impact on volumes of crimes in the year ending March 2023, it did find an impact on crime outcomes. So called ‘positive outcomes’ for violence/robbery (a measure of the number of crimes solved) increased by 8.1% in hot spots on days patrolled versus days not patrolled. Similar impacts were observed in other crime baskets, including all crimes minus possession, victim-based crimes and possession, with significant increases of 7.8%, 8.2% and 20.0% on patrol days, respectively. This suggests the programme is resulting in more robbery and violence offences being solved and more weapons and drugs being found in the hot spots. We found similar results for the year ending March 2022.
Cost-benefit analysis revealed that the social cost of the elements of the programme evaluated in this report for the year ending March 2023 was £17.5 million in financial year to March 2025 prices. It was not possible to monetise any benefits for the programme in that year, because there was no significant impact on offences at the programme level and, although there was an impact on positive outcomes, HO does not as yet have a method for monetising those impacts. This means we cannot say that the programme was cost effective in the year ending March 2023. However, it is worth noting that the total crime reduction benefits in the year ending March 2022 were estimated at £38.4 million, with programme costs totalling £16.8 million, in the financial year to March 2025 prices. That means overall, the programme’s crime reduction benefits (£38.4 million) still exceed the total cost across both years (£34.3 million).
4.2 Next steps
4.2.1 Future Grip
HO committed to delivering a further year of funding for the programme (up to year ending March 2024) for the same 20 PFAs to deliver targeted operational policing activity in violence hot spots. In the financial year to March 2025, the programme was broadened to include all forces and to also target ASB. Ongoing evaluation is also planned for that.
4.2.2 Alternative models
The evaluation model used in this analysis model compares crimes on patrol days versus crimes on non-patrol days (the crossover design). In some cases, the force planned a visitation pattern of visiting hot spots every day. This may be effective in reducing SV in high-crime areas but meant that it was not possible to calculate impact. In addition, it is also not possible to evaluate problem-solving using a day-based approach as it has a more permanent impact. Also, as the discussion section suggested, it may be that day-based models become less effective at capturing the impact of hot spots patrols over time, whereas this should not be the case for an area-based model.
For that reason, it would therefore be preferable to have other models that could calculate a robust effect in the hot spots by comparison with other similar areas. These models are therefore in development.
4.2.3 Residual deterrence
A limitation of the repeat crossover design used in this analysis is the potential for residual deterrence. That is, if treatment on a given day affects crime rates on subsequent unvisited control days, then the treatment effect will be diluted (Sherman, 2022). The wider evidence on residual deterrence is unclear, especially in UK studies (Barnes and others, 2020; Basford and others, 2021; Bland and others, 2021). Further testing is needed.
4.2.4 Problem-oriented policing
Most forces bgan running POP alongside hot spot policing in the year ending March 2023. We welcome this as the evidence for POP is strong. HO is working with forces to ensure their POP analytical setup is right, to enable analysis and evaluation of that activity.
4.2.5 Community insight
There is little evidence to show hot spot policing negatively impacts on police and community relations (Braga and others, 2019). However, the evidence is limited, and the potential impacts of hot spot policing on communities may depend on the approaches used and the context of the hot spots affected. To help understand more about the impact of Grip on individuals and communities, HO has created a separate fund for successful research proposals from forces.
Annex A: Technical Annex
This technical annex details the evaluation methodology for the crossover design using the DODO regression model. The aim of this work is to evaluate the effect of treatment on the crime rate within a hot spot area.
The analytical pipeline comprised the following stages:
1) Activity and crime data merge
The Home Office cleaned and formatted the activity data received from police forces so that each area and day was assigned a binary visitation flag, denoting ‘0’ - control or ‘1’ - treatment day. Treatment is defined as at least one additional scheduled visible officer patrol within the designated area.
HO crime data from the HODH were mapped to each area and aggregated to the area and day, matching the geospatial granularity of the activity data. All evaluations have been defined by either shapefiles – a bespoke area defined by the police force – or by an administrative boundary such as Output Area (OA), Lower Super Output Area (LSOA) or ward. Where HODH data was unavailable, the police force provided data independently.
For evaluations using the DODO model, crime baskets are listed at the end of this appendix. HO merged the processed activity and crime data to generate a record of treatment and crime count for each area and day.
Sometimes the offence location occurred directly on the boundary of a hot spot. This offence could be counted both in the hot spot and outside of it and could theoretically have led to double counting. This could have caused problems if the boundaries of hot spots were touching and therefore an offence could be counted in 2 hot spots. While uncommon as most forces left buffers around their hot spots, it did occur in a few locations.
2) Data processing and filters
In contrast to the previous report for the year ending March 2022, where the results were reported with and without filters, this report presents results without filters only.
In the last publication, 2 types of filters were applied to address operational challenges such as ramp-up periods, changes in experimental design and implementation (with periods of low or constant visitation) as well as low-crime areas. Following academic guidance and advice, we have dropped these filters.
Previously, the activity filter was designed to limit analysis to an appropriate period for each unit. The activity filter tested if a treated day was within a window between the first and last visited date extending 28 days (inclusive), forwards and backwards, which contained a minimum of 4 control days and 4 treatment days (note that the 4 control and treatment days did not have to come from the same 28-day window). HO then filtered the activity time series for the earliest and latest visited dates which met the above conditions, defining the analytical period.
The zero-crime day filter was designed to remove areas with a high proportion of days with zero crimes. This was due to their appropriateness as hot spots, and the ability to identify a treatment effect (Hinkle and others, 2013). Zero-crime days may be present because the area was poorly selected or because of intelligence that indicated the presence of organised crime, which may justify its visitation even if its recorded crime counts are suppressed/unreliable. The zero-crime day filter removed units with over 90% zero-crime days in their analytical period.
For the unfiltered model, the only restriction applied was to limit the period analysed for each unit between their first and last visited date. This was to mitigate contamination from other activities/operations.
3) DODO regression model
HO initially chose the negative binomial model as it has been used extensively in the evaluation of hot spot policing and the wider criminological literature to account for observed over-dispersion in crime count data (Braga and others, 2019; Weisburd and others, 2022). Following analytical quality assurance, it was recommended that a Poisson model was used instead of negative binomial, as it has been shown that the fixed effects Poisson model is robust to over-dispersion (Wooldridge, 1999). Testing the effect of the model with the current configuration, both the Poisson and negative binomial results with cluster robust standard errors were very similar and did not significantly impact the pooled treatment effect and associated CI.
The negative binomial and Poisson regressions were both implemented in R using the stats and MASS (Modern Applied Statistics with S) libraries, respectively. However, the results in this report used the Poisson regression models. The fixed effect regression model in this report used the following regression designs with a log-link function:
i) Effect of treatment on offences
crime count = treated + area + month + day + christmas + area:day + 1st of month [Equation 1]
Treated is a binary dummy variable indicating whether the area received a Grip patrol. Controls were also included for the area, month, day of the week (day), Christmas day and day-weekend interaction. The month and day controls were introduced to account for seasonality and within day variation in crime rates. The ‘1st of the month’ fixed effect was introduced to account for police-recorded crime recording effects, where an uncertain event date may be assigned to the start of the month. ‘Christmas day’ effect was added to control for the variation this day had on the forces. Treatment was assumed to be randomly assigned, conditional on controls such that there was no bias from unaccounted confounders.
ii) Effect of treatment on positive outcome volumes
outcome = treated + area + month + day + christmas + area:day + first of month [Equation 2]
The above regression model tests for the effect Grip patrols have had on outcome volumes. All independent variables remain the same as the first regression model.
iii) Effect of treatment on rate of positive outcomes
outcome rate = treated + area + month + day + christmas + area:day + first of month [Equation 3]
The rate is the volume of outcomes divided by the number of offences.
4) Crime reduction model
The effect size of visitation for each police force was collated in a spreadsheet model, together with associated p-values and estimate standard errors and CI. The change in the number of crimes due to visitation was estimated as:
∆_c = (N_v/(ε+1))-N_v [Equation 4]
Where N_v is the aggregate number of crimes on visited days and ε is the effect size. Hot spot areas and crimes which were filtered out of the analysis were assumed to have an effect size equal to the mean effect size across all shapefile or ward-level evaluations, as appropriate.
The estimated uncertainty associated with the effect size was propagated through Equation 2 using a Monte Carlo simulation, which randomly sampled the effect size of each PFA from a normal distribution characterised by the standard error estimate associated with the treatment regression coefficient. The central estimate and 95% CI on ∆_c were then reported.
Annex B: Updated year ending March 2022 results
Below are the updated and expanded year ending March 2022 Grip results. Similar to our year ending March 2023 results, these now include the estimated impact of patrols on offences, positive outcomes, and positive outcome rates in hot spots.
The estimated impact of the programme on SV and robbery differs from the original year ending March 2022 publication due to methodological improvements and updated data. These reasons include:
-
regression changes: as detailed in Annex A, several modifications have been made to the regression model – the model used in the year ending March 2022 publication controlled for weekend and area-weekend interactions whereas the updated model now incorporates day controls to account for daily variations in crime throughout the week
-
unfiltered results: the previous publication applied 2 types of filters that have now been removed – an activity filter that limited the analysis to periods with at least 4 activity/control days in every 28-day period; and a zero-crime day filter that excluded periods with a high percentage of days without offences
-
crime table updates: since the year ending March 2022 publication, police forces would have updated their crime tables for the period, either through additional offences, crime reclassifications or new outcomes; these updates are likely to cause limited changes to the previous results
Each section below only includes pooled estimates for the entire programme. Similar to the year ending March 2023 result section, impact results were produced on 6 crime baskets (see Annex G for a definition of each).
Offences:
The model still estimates a statistically significant reduction in violence/robbery in hot spots on patrol days. At the programme level, the model estimates a -5.2% reduction in SV on patrol days compared to non-patrol days (CI: -8.9% to -1.3%). This indicates that, on average and controlling for other factors, violence and robbery was 5.2% lower on patrol days compared to non-patrol days in the hot spots. This is slightly lower than the initial 7% estimate within the year ending March 2022 publication.
Note that this does not imply crime levels reduced within the hot spots; the analysis compares patrol days to non-patrol days within the same areas. Like the trend in the year ending March 2023, the model predicts a statistically significant increase in possession offences, with a 24.3% increase on patrol days within hot spots. There was no observed change in the number of offences for other crime baskets.
Table B.1: Forest plot and table of violence/robbery offences in the year ending March 2022
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 1 Design A, RCT | -48.23% | 4.23% | -22.00% | 0.100 | |
Police Force 2 RCT | -46.44% | 18.73% | -20.26% | 0.265 | |
Police Force 3 Design A | -28.17% | -0.63% | -14.40% | 0.040 | ** |
Police Force 6 Design A | -25.50% | 1.69% | -12.96% | 0.080 | * |
Police Force 5 RCT | -28.06% | 2.86% | -12.60% | 0.110 | |
Police Force 14 RCT | -24.70% | 4.03% | -11.49% | 0.139 | |
Police Force 12 RCT | -34.07% | 24.19% | -9.51% | 0.536 | |
Police Force 9 RCT | -17.71% | 2.91% | -7.40% | 0.160 | |
Police Force 10 RCT | -13.06% | 0.81% | -6.38% | 0.081 | * |
Pooled average | -8.63% | -1.27% | -5.17% | 0.011 | ** |
Police Force 7 | -19.02% | 11.05% | -5.17% | 0.510 | |
Police Force 3 Design B | -8.72% | -0.94% | -4.91% | 0.016 | ** |
Police Force 1 Design B | -10.69% | 3.43% | -3.89% | 0.289 | |
Police Force 12 | -12.61% | 8.77% | -2.50% | 0.650 | |
Police Force 6 Design B, RCT | -11.76% | 11.99% | -0.59% | 0.922 | |
Police Force 11 RCT | -8.24% | 10.09% | 0.51% | 0.913 | |
Police Force 8 | -8.90% | 11.15% | 0.63% | 0.901 | |
Police Force 15 | -7.19% | 9.26% | 0.70% | 0.866 | |
Police Force 16 | -6.00% | 14.70% | 3.83% | 0.458 | |
Police Force 4 | -18.19% | 42.47% | 7.96% | 0.588 | |
Police Force 13 | -5.26% | 28.77% | 10.45% | 0.204 |
Positive outcomes:
A statistically significant increase in positive outcomes was found for possession offences, with the model estimating a 50.2% increase (CI: 23.5% to 117.0%) on patrol days compared to non-patrol days in the hot spots. As possession offences also increased significantly, this suggests patrols are leading to more people being caught with weapons or drugs.
All other crime baskets experienced non-significant increases in positive outcomes on patrol days. Please note, the number of forces has decreased for positive outcomes compared to the violence/robbery result for the year ending March 2022 due to:
- forces supplying their own results
- forces supplying their own crime data
Table B.2: Forest plot and table of violence/robbery with a positive outcome in the year ending March 2022
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 15 | -43.62% | -8.39% | -28.13% | 0.008 | ** |
Police Force 1, Design B | -46.89% | 13.02% | -22.53% | 0.185 | |
Police Force 7 | -37.57% | 13.24% | -15.92% | 0.254 | |
Police Force 12 | -31.05% | 7.21% | -14.02% | 0.180 | |
Police Force 14 RCT | -42.41% | 42.86% | -9.29% | 0.674 | |
Police Force 16 | -18.08% | 26.56% | 1.82% | 0.871 | |
Police Force 11 RCT | -15.81% | 27.34% | 3.54% | 0.742 | |
Police Force 2 RCT | -65.05% | 219.69% | 5.71% | 0.922 | |
Police Force 8 | -13.50% | 30.55% | 6.27% | 0.563 | |
Pooled Average | -5.23% | 46.27% | 11.18% | 0.224 | |
Police Force 10 RCT | -5.08% | 33.29% | 12.48% | 0.174 | |
Police Force 3, Design B | 3.68% | 39.10% | 20.09% | 0.015 | ** |
Police Force 13 | -18.68% | 98.50% | 27.05% | 0.293 | |
Police Force 12 RCT | -58.32% | 500.755 | 58.24% | 0.500 | |
Police Force 4 | -8.24% | 190.265 | 63.20% | 0.095 | * |
Positive outcome rates:
There was a non-statistically significant increase in the rate of positive outcomes for violence/robbery offences on patrol days, with the model estimating a 26.1% rise (CI: -1.9% to 239.1%). The rate of positive outcomes is calculated by dividing the number of positive outcomes over the number of offences. This was significant at a 10% level and likely suggests that the rate of solving crime is higher on patrol days than non-patrol days in hot spots. This was; however, not observed in ASB which had a non-statistically significant increase in the rate of positive outcomes of 15.6% (CI: -9.2% to 128.8%).
Table B.3: Forest plot and table of positive outcome rate for violence/robbery offences in the year ending March 2022
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 1, Design B | -51.67% | 16.37% | -25.01% | 0.199 | |
Police Force 15 | -37.76% | -8.28% | -24.44% | 0.005 | ** |
Police Force 12 | -32.04% | 11.65% | -12.89% | 0.276 | |
Police Force 7 | -38.61% | 29.56% | -10.81% | 0.548 | |
Police Force 16 | -21.02% | 22.67% | -1.57% | 0.888 | |
Police Force 2 RCT | -86.66% | 661.29% | 0.79% | 0.994 | |
Police Force 8 | -14.78% | 31.90% | 6.02% | 0.600 | |
Police Force 11 RCT | -8.55% | 31.69% | 9.74% | 0.318 | |
Pooled Average | -1.94% | 239.09% | 26.06% | 0.081 | * |
Police Force 10 RCT | 4.32% | 52.39% | 26.08% | 0.016 | ** |
Police Force 3, Design B | 7.18% | 58.96% | 30.53% | 0.008 | *** |
Police Force 4 | -23.41% | 204.62% | 52.75% | 0.229 | |
Police Force 13 | 16.72% | 193.88% | 56.45% | 0.164 | |
Police Force 14 RCT | -15.60% | 212.28% | 62.35% | 0.147 | |
Police Force 12 RCT | -90.89% | 2886.51% | 64.98% | 0.735 |
Additional material relating to the year ending March 2022 displacement/diffusion results
The table below show the offence counts for the main hot spots and their buffer zones, for the forces used in the year ending March 2022 displacement/diffusion analysis.
Police force area | Sum of offences: Hot spot |
Sum of offences: Buffered zone |
---|---|---|
Police Force 1, Design B | 1441 | 308 |
Police Force 11, RCT | 1980 | 726 |
Police Force 14 RCT | 495 | 156 |
Police Force 2 RCT | 409 | 146 |
Police Force 4 | 240 | 166 |
Police Force 6, Design C | 1827 | 948 |
Police Force 7 | 1794 | 377 |
Annex C: Force-level delivery descriptions
Avon and Somerset
Avon and Somerset ran 15-minute patrols in 34 hexagonal hot spots with a height of 200m. Patrols were randomly allocated by day with 4 in every 7 days to receive patrols on average (57% target coverage). The patrols were largely tasked over time and followed a non-randomised pattern and targeted times of the day when crime was highest. The top 15 hot spots accounted for 5% of all non-domestic violence against the person in the force. A GPS tracking system was adopted for the first part of the year but was later dropped due to inaccuracies and data collection reverted to manual methods, verified by an inspector watching (dip-sampled) footage from body-worn video.
Bedfordshire
Bedfordshire funded several different hot spot operations with their funding. The main one ran in 22 LSOAs. Patrols were not randomly allocated. The patrol schedule was for at least one 15-minute patrol between 2pm and 10pm, at least once every 3 days. However, in practice, much higher levels of patrolling were delivered. This was GPS-tracked through a third-party app. The other funded operations, which were not GPS-tracked, targeted night-time economy areas or larger ward-level hot spots. Data for these could only be returned at ward level so they were not included as part of the evaluation.
Cleveland
Cleveland patrolled 32 hexagonal hot spots with sides of 200m. The patrol schedule was for each hot spot to be treated every other day for 15 to 20 minutes. Patrols were carried out by police officers and police community support officers on overtime. Patrols were tracked using Airwave radios from August 2022 to July 2023.
Essex
Essex ran patrols into 35 hexagonal hot spots with heights of 200m. The aim was to patrol each hot spot for at least 15 minutes every day and to test the impact against a set of (non-randomly) selected control areas. However, compliance for April to December was around 50%, increasing markedly in January to March to around 90%. In addition, testing suggested the selected control areas had lower levels of crime than the intervention areas. The decision was therefore taken to run a day-based comparison model (given the compliance rates allowed for a high enough number of control days) rather than an area-based comparison model. Essex used officer-completed forms to capture patrols and this was then checked using GPS data from Airwaves radios.
Greater Manchester Police (GMP)
GMP originally mapped more than 100 bespoke-shaped hot spots (average area 286,549m2) for the year ending March 2023. Of these, they ended up patrolling 85 to a level sufficient for analysis. They were generally aiming for one 15 to 20-minute patrol every 3 days across the hot spots (non-randomised), with a few of the highest crime areas receiving more. GMP also began problem-solving activity in the year ending March 2023. This resulted in some additional patrol activity, which was captured in their data returns. This was added into their main patrol activity for the purposes of our crime-impact estimation. GMP patrols were partly tasked through a central team (in a subset of hot spots), and partly through district neighbourhood policing teams. GMP captured their patrol activity manually in the year ending March 2023 but worked to develop their own GPS solution through the year, which they began using the following year.
Hampshire
Hampshire switched from LSOA-sized hot spots to 15m x 250m-sided boxes in January 2022. These hot spots were largely retained throughout the year ending March 2023, with a few substituted in and out each quarter, meaning that 25 were patrolled in total during that period. For that reason, Hampshire’s year ending March 2023 crime-impact result covered the period from January 2022 through to March 2023. Hampshire’s patrol schedule was quasi-random in that they aimed to patrol each hot spot every other day. They were largely successful in this regard, with compliance rates (for a patrol of any length) of above 90% most quarters. Hampshire collected their activity data manually, although they were exploring a GPS solution for use the following year.
Humberside
Throughout the year ending March 2023, Humberside continued to use the same 98 bespoke-shaped hot spots that they began patrolling the previous year (average area 45,800m2). The patrol schedule was for 52 of the 98 hot spots to be randomly selected for at least 15 minutes patrolling each day (implying overall coverage on 53% of days). Compliance with this schedule was good: above 90% most quarters, for a patrol of any length. The force tracked compliance with a third-party GPS app.
Kent
Kent patrolled 31 hot spots in the year ending March 2023 that were either single LSOAs or groups of LSOAs. Officers were briefed on the specific micro-locations within these areas that warranted the most patrolling attention. Given the large size of the hot spots, Kent also used longer patrols than most forces, with officers sometimes spending a full 8-hour shift in the hot spots, aiming for multiple 15-minute patrols of the key micro-locations. Kent aimed for a non-random schedule in which each hot spot was patrolled every third day on average. Compliance was mixed with low rates in some hot spots. Kent did not use GPS tracking and manually completed data returns using a QR-code system. Kent also began problem-solving activity in the year ending March 2023. This resulted in some additional patrol activity, which was captured in their data returns. This was added into their main patrol activity for the purposes of our crime-impact estimation.
Lancashire
Lancashire ran patrols in 21 hot spots at the LSOA level. They refreshed these halfway through the year, substituting some out for new LSOAs. In total, 30 LSOAs were patrolled at some point during the year. They used a randomised schedule of patrol days, aiming for 15 to 30-minute visits with 2 officers on about 40% to 45% of days. Compliance was around 60% to 70%, meaning areas typically received a patrol every fourth day on average. Officers were tasked mostly via overtime. Lancashire used a third-party GPS app to track patrols.
Leicestershire
Leicestershire ran patrols in 27 hot spots from August 2022 to the end of June 2023. These were square with sides of 150m. They used a randomised schedule of blocks of 3 days with 3 days off in 9. However, compliance was around 50%, meaning that areas received a patrol about one day in 3 on average. Some hot spots were patrolled using a dedicated central team, which some were tasked using neighbourhood officers. They aimed for 15 minutes of patrolling per visit, with some visits lasting up to an hour. The force purchased GPS trackers for the patrols but had to manually enter that data for part of the year before generating an automated solution towards the end of the year.
Merseyside
Merseyside had 12 main hot spots in the year ending March 2023 that were bespoke shapes with average area 311,063m2. They also had 2 larger areas for night-time economy patrols. One of these over-lapped some of the other hot spots, so was excluded from the analysis. Even so, it is possible that patrols for that area may have been done on control days for the overlapping hot spots which could ‘contaminate’ Merseyside’s crossover design and that needs to be borne in mind when interpreting results. Merseyside ran 2 RCTs in the year ending March 2023. For both RCTs, Merseyside randomised blocks of 3 to 5 treatment days, followed by blocks of 3 to 5 control days. Each randomised schedule is then assigned to a hot spot. One RCT was funded using central Grip funding, while the other was funded via the matched funding requirement using duty police officers. Merseyside typically patrolled using one sergeant and 4 to 5 constables, to enable visibility to continue if some officers were abstracted for other incidents. Merseyside task patrols in 9 of the hot spots via overtime with the rest tasked by duty staff. Merseyside also began problem-solving activity in the year ending March 2023. This resulted in some additional patrol activity, which was captured in their data returns. This was added into their main patrol activity for the purposes of our crime-impact estimation. Merseyside tracked their patrols using GPS data from their Airwaves radio system.
Metropolitan Police Service (MPS)
MPS targeted 75 hexagonal hot spots with 200m sides (103,923m2). MPS aimed to switch mainly to problem-solving activity in the year ending March 2023, while maintaining a baseline level of patrolling in the treatment hot spots (at least 2 days in 7). Each of the 75 areas had their own POP champion. However, the level of problem-solving was variable across hot spots, with some areas not reaching the response phase of the SARA (scanning, analysis, response, assessment) plan[footnote 1] until the fourth quarter (January to March), which needs to be borne in mind when interpreting results. The MPS used an internal GPS tracking system in the year ending March 2022, but this system was upgraded during the year ending March 2023. This meant that data was not returned to HO at hot spot level for much of the year. Therefore, it was not possible to judge patrolling coverage or compliance at the hot spot level.
Northumbria
Northumbria has 2 sets of hot spot areas running from April 2022 to June 2023. There were 25 main hot spots (average 2,353,390m2) and 2 much larger city centre areas, all bespoke-shaped. Their patrol schedule was quasi-random, in that they attempted at least one 60-minute patrol every other day, except for the city centre areas where they aimed for 4-hour patrols. This was largely achieved – compliance rates were over 90% for most quarters, for patrols of any length. These hot spots were used until the end of June 2023, so the crime impact for the year ending March 2023 actually includes an additional quarter of activity. The force tracked all patrols using a third-party GPS app (swapped some out and added some)
Nottinghamshire (Notts)
They patrolled about 17 hot spots each quarter but substituted several in and out, such that 26 hot spots were patrolled in total. The hot spots were the standard police beats of varying size but up to several square kilometres. Notts used a mixture of dedicated central teams and overtime to task patrols. They did not use GPS tracking, so officer-completed returns provided the data. Notts aimed to patrol hot spots 3 times a week. Patrol length varied by hot spot, with the range of intended patrol length being 1 to 8 hours to ensure good visibility across the hot spot. Notts also began problem-solving activity in the year ending March 2023. This resulted in some additional patrol activity, which was captured in their data returns. This was added into their main patrol activity for the purposes of our crime-impact estimation.
South Wales
South Wales returned patrol data to HO at ward level. Originally, they aimed for patrol ‘treatment’ of 3 to 6 days, followed by 4 days off. However, discussions revealed that South Wales had not quite been following the core Grip model of identifying persistent long-term hot spots and developing a proactive patrol schedule for those areas. Instead, they tended to be more reactive, with local commanders determining hot spots and patrol schedules on a more ad hoc basis. Because of that, they were excluded from the analysis. South Wales also did not have GPS tracking for the year ending March 2023. However, they worked to on-board one through the year and hoped to have that in place for the year ending March 2024.
South Yorkshire (SYP)
SYP ran patrols into 60 hot spots from July 2022 until the end of March 2023. The hot spots were hexagonal with an area of 20,000m2. Patrols were randomised by day, with 30 of the hot spots receiving a 15-minute foot patrol and 30 acting as controls each day. Compliance was extremely high – well over 90%. SYP used purchased stand-alone GPS devices to track their Grip patrols, combined with mapping software to enable the accurate mapping and analysis of patrol activity.
Sussex
Sussex ran patrols into 15 hot spots which were roughly 150m x 150m but drawn to reflect local geography. Sussex ran an RCT from 1 January to 30 April 2022. This was not recorded in the original year ending March 2022 report, but its impact has now been captured and included in this report (in the year ending March 2022 meta-analysis, see Annex B). In the year ending March 2023, Sussex randomised patrol days in a subset of hot spots. For the rest of the year ending March 2023 patrols, Sussex used a quasi-random patrol schedule of every other day for 15 to 20 minutes. Sussex also began problem-solving activity in the year ending March 2023. This resulted in some additional non-randomised patrol activity, which was captured in their data returns. This was added into their main patrol activity for the purposes of our crime-impact estimation. They used handheld GPS devices to track the patrols. Sussex tasked the patrols to their hot spots using both police and business wardens.
Thames Valley Police (TVP)
TVP had 34 bespoke-shaped hot spots in the year ending March 2023 (average size 234,842m2) each with 2 specific patrol areas within them. They tasked patrolling via their own GPS tracking app, which randomly selected half the areas to be patrolled each day for at least 15 minutes. TVP changed their tasking approach from a dedicated team in the year ending March 2022 to filling patrols from business-as-usual tasking in the year ending March 2023. They also removed the element of officer choice that had caused variable compliance in the year ending March 2022 and briefed officers on the hottest times to patrol each hot spot. However, the move to business-as-usual tasking did cause a marked increase in vehicle patrols being used instead of foot patrols (about 80% in some quarters). TVP aimed to correct this in the year ending March 2024.
West Midlands
West Midlands targeted 57 bespoke-shaped hot spots in the year ending March 2023, with an average size of 218,896m2. They used a patrol schedule randomised by day with hot spots targeted to receive 15 to 45 minute patrols on 50% of days on average. However, compliance was around 60% to 70% most quarters, meaning that areas received a patrol about one day in 3. For most of the year ending March 2023, West Midlands used officer forms to capture attendance and time in the hot spots, checked using Airwaves data. However, in the final months of the year ending March 2023, the force began testing their own GPS app to be used to track all patrols in the year ending March 2024.
West Yorkshire Police (WYP)
WYP targeted 52 hot spots, most of which were single hexagons with 125m sides. However, a few of the hot spots were larger areas of multiple hexagons joined together. The latter were targeted with 45-minute patrols of 6 officers, while the single hexagons received 20-minute patrols with 2 officers. Patrols were randomised by day into blocks of 4 days on, 4 days off, meaning that the target was for each area to be patrolled on 50% of days on average. Compliance was high – generally above 90% for a patrol of any length. WYP also began problem-solving activity in the year ending March 2023. This resulted in some additional patrol activity, which was captured in their data returns. This was added into their main patrol activity for the purposes of our crime-impact estimation. WYP used both officer reports and a GPS tracking system to record patrols, and generally these showed a high level of alignment. The tracking system used Airwaves data that were processed and returned to WYP by a third-party contractor.
Annex D: Alternative meta-analysis
The main overall result reported in the publication is using the mean average of the effect size across each evaluation. This implicitly weights each evaluation equally. An alternative approach in the meta-analysis literature weights the evaluations by the inverse variance with random effects. The table below summarises the results of this approach, showing a statistically significant result for possession offences only.
Overall pooling | SV | ASB | All crimes minus possession | NC | Possession offences |
---|---|---|---|---|---|
Effect size | 1.03% | 1.52% | 0.30% | 0.97% | 14.69% |
CI Lower | -1.02% | -0.57% | -0.76% | -1.87% | 9.42% |
CI Higher | 3.11% | 3.65% | 1.37% | 3.89% | 20.22% |
Significance | 0.326 | 0.155 | 0.582 | 0.507 | 0.000 |
Applying this methodology to positive outcomes gave statistically significant results at the 0.1 level for SV offences and at the 0.05 level for possession offences, victim-based offences and all crimes minus possession.
Annex E: Innovation Fund delivery description
Police force impact results are anonymised throughout this report. Innovation Fund design choices have therefore been anonymised to enable us to retain anonymity when discussing operational designs of Innovation Fund interventions with significant results in Section2.3.
Police Force 6:
This force trialled 2 approaches: fixed-post hot spots policing at hot street junctions and the use of drones in a highly visible manner in serious violence/robbery hot spots. Street junctions were randomly chosen from an eligible list, and randomly assigned days for fixed-post presence. Drones were deployed in a randomised selection of existing SV/robbery hot spots in an attempt to increase visible presence and provide an enhanced CCTV function.
Police Force 10:
This force trialled a 2-strand approach to respond to public concern regarding spiking, and to respond to violence against women and girls in night-time economy areas. The approach involved deploying plain-clothes observers to target potential offenders in the night-time economy to prevent predatory behaviour and prevent sexual assaults. Uniformed officers also patrolled to intervene where appropriate. The second strand of the approach involved deploying volunteer guardians, who could provide support and advice to people in the night-time economy, especially women and girls.
Police Force 12:
The project employs additional door supervisors or ‘marshals’ on a Friday and Saturday night between 20:00 hours to 04:00 hours. These staff complement existing police resources, undertaking high visibility foot patrols and engagement within the night-time economy. The marshal project’s ‘innovative’ element is the collaborative, partnership approach to tackling SV and violence against women and girls.
Police Force 16:
This Innovation Fund project funded regular street outreach activity in SV hot spots focusing particularly on diverting young people and those suffering with substance misuse. Street outreach activity comprised foot patrols in 8 SV hot spots between peak hours for youth violence, in addition to policing activity in those hot spots.
Police Force 19:
This project had several strands, including the combined use of uniformed and plain-clothes patrols to spot subjects acting suspiciously within the hot spots and engaging with them to prevent crime. This was combined with a social media campaign that targeted advertisements at the geofenced postcodes in and around the hot spots, giving prevention messages to people aged 16 to 30 years. The project focused on a range of crime types, including SV, violence against women and girls, drugs offences and public order offences.
Annex F: Full results for year ending March 2023
Below are the entire Grip results for the year ending March 2023, which were not included in the main body. These include the estimated impact of patrols on offences, positive outcomes and positive outcome rates in hot spots.
Each section includes a pooled estimate for the entire programme and force breakdown. Neighbourhood Crimes (NC) positive outcomes are not included due to a low sample size. Offences
Forest plot and table of ASB offences in the year ending March 2023. While the overall meta-analysis was not significant, 2 specific police forces demonstrated significant results: a significant reduction in Police Force 5 and a positive increase in Police Force 19.
Table F.1: Forest plot and table of ASB offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 5 | -34.44% | -0.08% | -19.06% | 0.049 | ** |
Police Force 11 | -24.10% | 9.22% | -8.95% | 0.312 | |
Police Force 10 | -14.25% | 9.31% | -3.19% | 0.601 | |
Police Force 7 | -13.12% | 9.80% | -2.33% | 0.963 | |
Police Force 15 | -9.53% | 5.72% | -2.20% | 0.576 | |
Police Force 14 | -10.20% | 7.58% | -1.71% | 0.708 | |
Police Force 1 | -4.59% | 3.58% | -0.59% | 0.778 | |
Police Force 20 | -6.47% | 8.61% | 0.78% | 0.838 | |
Police Force 13 | -7.62% | 11.70% | 1.59% | 0.745 | |
Police Force 9 | -8.42% | 13.81% | 2.09% | 0.709 | |
Pooled average | -1.82% | 13.08% | 3.48% | 0.250 | |
Police Force 17 | -8.78% | 17.52% | 3.54% | 0.590 | |
Police Force 2 | -4.65% | 12.58% | 3.61% | 0.403 | |
Police Force 18 | -4.68% | 12.80% | 3.69% | 0.399 | |
Police Force 3 | -5.07% | 14.56% | 4.28% | 0.382 | |
Police Force 4 | -4.43% | 15.86% | 5.22% | 0.300 | |
Police Force 12 | -1.88% | 18.16% | 7.67% | 0.119 | |
Police Force 19 | 3.13% | 19.82% | 11.16% | 0.006 | *** |
Police Force 16 | -29.03% | 227.27% | 52.41% | 0.280 |
Forest plot and table of NC offences in the year ending March 2023. The meta-analysis was non-significant but 5 individual forces demonstrated significant changes, with 3 experiencing reductions and 2 showing increases in NC on patrol days.
Table F.2: Forest plot and table of NC offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 14 | -48.38% | -0.69% | -28.40% | 0.045 | ** |
Police Force 10 | -24.72% | 0.13% | -13.18% | 0.052 | * |
Police Force 11 | -29.94% | 8.38% | -12.86% | 0.216 | |
Police Force 7 | -23.09% | 4.16% | -10.50% | 0.152 | |
Police Force 13 | -21.46% | 4.52% | -9.40% | 0.176 | |
Police Force 3 | -16.97% | 1.07% | -8.39% | 0.080 | * |
Police Force 18 | -17.66% | 8.61% | -5.43% | 0.429 | |
Pooled average | -3.91% | 4.75% | 0.23% | 0.921 | |
Police Force 5 | -28.37% | 43.70% | 1.46% | 0.935 | |
Police Force 20 | -5.53% | 9.24% | 1.59% | 0.670 | |
Police Force 9 | -7.53% | 15.43% | 3.32% | 0.564 | |
Police Force 1 | -2.67% | 10.17% | 3.55% | 0.270 | |
Police Force 17 | -17.00% | 29.97% | 3.86% | 0.740 | |
Police Force 15 | -5.09% | 16.55% | 5.17% | 0.336 | |
Police Force 12 | -13.16% | 27.52% | 5.24% | 0.602 | |
Police Force 4 | -12.57% | 27.62% | 5.63% | 0.570 | |
Police Force 2 | -4.19% | 16.84% | 5.81% | 0.265 | |
Police Force 19 | 6.64% | 34.16% | 19.61% | 0.002 | *** |
Police Force 16 | 0.37% | 70.09% | 30.66% | 0.047 | ** |
Forest plot and table of all crimes minus possession offences in the year ending March 2023. While the overall analysis yielded non-significant results, Police Force 19 demonstrated a significant positive increase at the 10% significance level.
Table F.3: Forest plot and table of all crimes minus possession offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 11 | -12.86% | 4.30% | -4.76% | 0.298 | |
Police Force 10 | -7.75% | 2.40% | -2.80% | 0.286 | |
Police Force 7 | -8.32% | 5.23% | -1.78% | 0.610 | |
Police Force 5 | -17.71% | 17.48% | -1.68% | 0.852 | |
Police Force 4 | -6.64% | 3.79% | -1.56% | 0.559 | |
Police Force 14 | -6.63% | 4.43% | -1.25% | 0.659 | |
Police Force 9 | -5.71% | 3.73% | -1.10% | 0.649 | |
Police Force 1 | -3.03% | 1.23% | -0.92% | 0.398 | |
Police Force 20 | -2.84% | 3.54% | 0.30% | 0.855 | |
Police Force 3 | -3.83% | 4.77% | 0.37% | 0.864 | |
Police Force 13 | -4.51% | 6.46% | 0.83% | 0.767 | |
Poooled average | -1.21% | 3.40% | 0.97% | 0.399 | |
Police Force 2 | -1.93% | 5.05% | 1.50% | 0.395 | |
Police Force 15 | -2.50% | 6.56% | 1.93% | 0.398 | |
Police Force 12 | -5.89% | 10.47% | 1.96% | 0.634 | |
Police Force 18 | -1.53% | 7.90% | 3.08% | 0.194 | |
Police Force 19 | -0.44% | 7.07% | 3.25% | 0.085 | * |
Police Force 17 | -3.05% | 10.54% | 3.52% | 0.301 | |
Police Force 16 | -12.14% | 50.05% | 14.82% | 0.312 |
Forest plot and table of victim-based offences in the year ending March 2023. The meta-analysis and force-level results were non-significant.
Table F.4: Forest plot and table of victim-based offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 11 | -15.73% | 1.91% | -7.33% | 0.116 | |
Police Force 4 | -9.47% | 2.92% | -3.47% | 0.280 | |
Police Force 10 | -6.56% | 2.76% | -2.01% | 0.403 | |
Police Force 3 | -6.35% | 3.01% | -1.78% | 0.460 | |
Police Force 20 | -4.18% | 2.39% | -0.95% | 0.573 | |
Police Force 1 | -2.78% | 1.48% | -0.67% | 0.538 | |
Police Force 14 | -7.13% | 6.72% | -0.44% | 0.900 | |
Police Force 9 | -5.28% | 4.74% | -0.40% | 0.876 | |
Police Force 7 | -8.15% | 8.77% | -0.05% | 0.991 | |
Pooled average | -1.43% | 2.16% | 0.31% | 0.741 | |
Police Force 13 | -4.71% | 6.91% | 0.94% | 0.751 | |
Police Force 5 | -15.70% | 21.37% | 1.15% | 0.902 | |
Police Force 18 | -2.80% | 7.16% | 2.06% | 0.413 | |
Police Force 19 | -2.16% | 6.59% | 2.12% | 0.337 | |
Police Force 2 | -2.36% | 7.19% | 2.30% | 0.339 | |
Police Force 12 | -7.10 | 13.30% | 2.59% | 0.613 | |
Police Force 16 | -7.53% | 14.42% | 2.86% | 0.604 | |
Police Force 15 | -2.02% | 9.15% | 3.41% | 0.223 | |
Police Force 17 | -3.88% | 12.93% | 4.19% | 0.318 |
Positive outcomes:
Forest plot and table of ASB offences with a positive outcome in the year ending March 2023. The overall meta-analysis was significant at a 10% level, with an estimated increase of 6.22%.
Table F.5: Forest plot and table of ASB offences with a positive outcome, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 5 | -60.35% | 43.28% | -24.63% | 0.388 | |
Police Force 3 | -33.88% | 26.68% | -8.48% | 0.593 | |
Police Force 2 | -20.73% | 5.82% | -8.41% | 0.233 | |
Police Force 13 | -34.92% | 30.07% | -7.99% | 0.637 | |
Police Force 14 | -25.77% | 22.41% | -4.68% | 0.707 | |
Police Force 4 | -17.59% | 22.05% | 0.29% | 0.977 | |
Police Force 1 | -6.39% | 11.91% | 2.35% | 0.610 | |
Police Force 18 | -8.98% | 17.23% | 3.30% | 0.615 | |
Police Force 12 | -21.77% | 39.72% | 4.55% | 0.764 | |
Police Force 10 | -15.45% | 29.47% | 4.62% | 0.678 | |
Pooled Average | -0.60% | 13.89% | 6.22% | 0.077 | * |
Police Force 17 | -28.10% | 71.90% | 11.18% | 0.634 | |
Police Force 20 | -16.80% | 49.74% | 11.62% | 0.463 | |
Police Force 15 | -6.82% | 38.23% | 13.49% | 0.208 | |
Police Force 16 | -12.59% | 51.24% | 14.97% | 0.318 | |
Police Force 9 | -4.66% | 40.56% | 15.77% | 0.139 | |
Police Force 11 | -16.50% | 67.27% | 18.18% | 0.346 | |
Police Force 7 | -5.05% | 60.30% | 23.37% | 0.116 | |
Police Force 19 | -5.24% | 60.84% | 23.46% | 0.118 |
Forest plot and table of all crimes minus possession offences, with a positive outcome in the year ending March 2023. The meta-analysis revealed a statistically significant effect of 7.76%.
Table F.6: Forest plot and table of all crimes minus possession offences with a positive outcome, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 13 | -26.48% | 15.98% | -7.66% | 0.493 | |
Police Force 1 | -9.71% | 2.96% | -3.58% | 0.276 | |
Police Force 14 | -16.23% | 11.22% | -3.48% | 0.625 | |
Police Force 2 | -11.40% | 13.34% | 0.21% | 0.973 | |
Police Force 20 | -6.43% | 15.59% | 4.00% | 0.467 | |
Police Force 10 | -8.24% | 17.97% | 4.04% | 0.536 | |
Police Force 3 | -10.52% | 24.77% | 5.67% | 0.516 | |
Police Force 9 | -4.85% | 18.06% | 5.99% | 0.290 | |
Police Force 11 | -10.33% | 27.34% | 6.85% | 0.459 | |
Pooled Average | 3.69% | 12.35% | 7.76% | 0.000 | ** |
Police Force 7 | -5.29% | 23.12% | 7.99% | 0.251 | |
Police Force 18 | -0.78% | 19.10% | 8.71% | 0.073 | * |
Police Force 4 | -4.57% | 26.14% | 9.72% | 0.193 | |
Police Force 16 | -8.36% | 32.80% | 10.31% | 0.300 | |
Police Force 15 | 3.20% | 19.41% | 11.01% | 0.005 | ** |
Police Force 12 | -4.38% | 29.90% | 11.45% | 0.165 | |
Police Force 19 | 1.76% | 24.40% | 12.51% | 0.021 | ** |
Police Force 17 | -7.74% | 39.43% | 13.42% | 0.232 | |
Police Force 5 | -2.41% | 92.31% | 36.99% | 0.069 | * |
Forest plot and table of victim-based offences with a positive outcome in the year ending March 2023. The meta-analysis revealed a statistically significant effect of 8.23% on patrol days.
Table F.7: Forest plot and table of all crimes minus possession offences with a positive outcome, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 11 | -23.19% | 15.52% | -5.80% | 0.566 | |
Police Force 14 | -19.16% | 15.30% | -3.46% | 0.698 | |
Police Force 1 | -12.26% | 6.53% | -3.32% | 0.495 | |
Police Force 13 | -23.95% | 23.17% | -3.21% | 0.791 | |
Police Force 16 | -21.76% | 25.37% | -0.96% | 0.936 | |
Police Force 20 | -8.08% | 18.20% | 4.23% | 0.518 | |
Police Force 3 | -11.24% | 23.09% | 4.53% | 0.596 | |
Police Force 10 | -7.35% | 18.09% | 4.60% | 0.467 | |
Police Force 9 | -5.35% | 19.92% | 6.54% | 0.294 | |
Pooled Average | 3.24% | 14.31% | 8.23% | 0.000 | ** |
Police Force 7 | -7.86% | 27.82% | 8.52% | 0.327 | |
Police Force 2 | -5.52% | 26.92% | 9.51% | 0.228 | |
Police Force 19 | -1.24% | 22.90% | 10.17% | 0.083 | * |
Police Force 12 | -8.01% | 34.52% | 11.24% | 0.272 | |
Police Force 18 | -0.35% | 26.11% | 12.10% | 0.057 | * |
Police Force 4 | -3.09% | 30.78% | 12.58% | 0.121 | |
Police Force 15 | 2.42% | 25.65% | 13.44% | 0.016 | ** |
Police Force 17 | -4.88% | 40.33% | 15.54% | 0.145 | |
Police Force 5 | -10.14% | 131.61% | 44.27% | 0.129 |
####Positive outcome rate:
Forest plot and table of positive outcome rate for victim-based offences in the year ending March 2023. The meta-analysis found a significant increase in the rate of positive outcomes for victim-based crimes of 8.8%. This means the rate of solving victim-based crimes statistically increased on patrol days.
Table F.8: Forest plot and table of positive outcome rate for victim-based offences, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 13 | -26.78% | 21.75% | -5.58% | 0.658 | |
Police Force 14 | -17.01% | 18.42% | -0.87% | 0.924 | |
Police Force 11 | -15.69% | 17.29% | -0.55% | 0.947 | |
Police Force 1 | -12.33% | 13.39% | -0.29% | 0.964 | |
Police Force 16 | -21.07% | 28.40% | 0.67% | 0.957 | |
Police Force 20 | -12.19% | 19.72% | 2.53% | 0.752 | |
Police Force 17 | -13.99% | 25.52% | 3.90% | 0.691 | |
Police Force 18 | -9.66% | 19.66% | 3.97% | 0.587 | |
Police Force 7 | -14.03% | 26.53% | 4.30% | 0.669 | |
Police Force 12 | -9.15% | 21.80% | 5.19% | 0.499 | |
Police Force 2 | -12.99% | 28.60% | 5.78% | 0.573 | |
Police Force 3 | -12.27% | 27.59% | 5.80% | 0.555 | |
Police Force 15 | -4.99% | 17.84% | 5.81% | 0.304 | |
Police Force 9 | -7.83% | 27.43% | 8.37% | 0.330 | |
Pooled Average | 3.23% | 15.93% | 8.84% | 0.001 | ** |
Police Force 10 | -3.89% | 24.26% | 9.28% | 0.176 | |
Police Force 19 | -3.70% | 26.18% | 10.23% | 0.158 | |
Police Force 4 | -2.70% | 33.92% | 14.15% | 0.104 | |
Police Force 5 | 10.37% | 185.75% | 77.59% | 0.018 | ** |
Forest plot and table of positive outcome rate for all crimes minus possession in the year ending March 2023. Like the above, the meta-analysis was significant and estimated to be 8.3%.
Table F.9: Forest plot and table of positive outcome rate for all crimes minus possession, year ending March 2023
Police force area | CI Lower | CI Higher | Effect | P-value | Sig |
---|---|---|---|---|---|
Police Force 13 | -25.79% | 13.37% | -8.27% | 0.424 | |
Police Force 2 | -17.72% | 12.05% | -3.98% | 0.606 | |
Police Force 1 | -6.87% | 9.48% | 0.98% | 0.814 | |
Police Force 14 | -11.35% | 16.40% | 1.58% | 0.821 | |
Police Force 17 | -17.99% | 27.04% | 2.07% | 0.854 | |
Police Force 15 | -5.57% | 11.06% | 2.41% | 0.565 | |
Police Force 20 | -11.82% | 18.95% | 2.42% | 0.754 | |
Police Force 18 | -8.15% | 17.57% | 3.92% | 0.542 | |
Police Force 12 | -10.22% | 21.74% | 4.54% | 0.567 | |
Police Force 7 | -10.54% | 23.69% | 5.19% | 0.540 | |
Police Force 9 | -6.81% | 24.01% | 7.50% | 0.321 | |
Pooled Average | 3.49% | 14.03% | 8.31% | 0.000 | ** |
Police Force 3 | -10.50% | 31.71% | 8.57% | 0.404 | |
Police Force 10 | -4.12% | 25.47% | 9.68% | 0.178 | |
Police Force 19 | -2.96% | 24.45% | 9.89% | 0.137 | |
Police Force 4 | -6.42% | 29.39% | 10.04% | 0.247 | |
Police Force 11 | -3.59% | 30.79% | 12.29% | 0.136 | |
Police Force 16 | -6.13% | 43.51% | 16.07% | 0.169 | |
Police Force 5 | 4.91% | 136.89% | 57.65% | 0.028 | ** |
Annex G: Crime baskets
List of offences for crime baskets and the definition of positive outcomes.
Violence and robbery:
-
Assault with injury
-
Assault with intent to cause serious harm
-
Assault without injury
-
Assaults on emergency workers (other than constables)
-
Attempted murder
-
Endangering life
-
Manslaughter
-
Murder
-
Racially or religiously aggravated assault with injury
-
Racially or religiously aggravated assault without injury
-
Robbery of business property
-
Robbery of personal property
-
Threats to kill
Assaults on constables were not included in the crime count as these can only occur if there is a police presence. We also exclude all stalking and harassment offences as these tend to occur over multiple locations and hence are hard to precisely geo-locate to hot spots.
ASB offences:
-
Arson endangering life
-
Arson not endangering life
-
Criminal damage to a building other than a dwelling
-
Criminal damage to a dwelling
-
Criminal damage to a vehicle
-
Other criminal damage
-
Racially or Religiously aggravated criminal damage
-
Trafficking in controlled drugs
-
Other offences against the state and public order
-
Public fear alarm or distress
-
Racially or Religiously aggravated public fear alarm or distress
-
Violent disorder
Possession offences:
-
Possession of firearms offences
-
Possession of other weapons
-
Possession of firearms with intent
-
Other firearms offences
-
Possession of controlled drugs (cannabis)
-
Possession of controlled drugs excluding cannabis
-
Other drug offences
-
Possession of article with blade or point
Neighbourhood crime offences:
-
Aggravated burglary – residential
-
Aggravated burglary in a dwelling
-
Aggravated vehicle taking
-
Attempted burglary – residential
-
Attempted burglary in a dwelling
-
Attempted distraction burglary – residential
-
Attempted distraction burglary in a dwelling
-
Burglary – residential
-
Burglary in a dwelling
-
Distraction burglary - residential
-
Distraction burglary in a dwelling
-
Interfering with a motor vehicle
-
Robbery of personal property
-
Theft from a vehicle
-
Theft from the person
-
Theft or unauthorised taking of a motor vehicle
Victim-based offences:
-
Abuse of children through sexual exploitation
-
Abuse of position of trust of a sexual nature
-
Aggravated vehicle taking
-
Arson endangering life
-
Arson not endangering life
-
Assault with injury
-
Assault with injury on a constable
-
Assault with intent to cause serious harm
-
Assault without injury
-
Assault without injury on a constable
-
Assaults on emergency workers (other than constables)
-
Attempted murder
-
Blackmail
-
Causing death by careless driving when under the influence of drink or drugs
-
Causing death by careless or inconsiderate driving
-
Causing death or serious injury by dangerous driving
-
Causing death or serious injury by driving: unlicensed disqualified or uninsured drivers
-
Causing sexual activity without consent
-
Child abduction
-
Conspiracy to murder
-
Controlling and coercive behaviour
-
Corporate manslaughter
-
Criminal damage to a building other than a dwelling
-
Criminal damage to a dwelling
-
Criminal damage to a vehicle
-
Cruelty to children/young persons
-
Dishonest use of electricity
-
Endangering life
-
Exposure and voyeurism
-
Harassment
-
Incest or familial sexual offences
-
Intentional destruction of a viable unborn child
-
Interfering with a motor vehicle
-
Kidnapping
-
Malicious communications
-
Manslaughter
-
Modern slavery
-
Murder
-
Other criminal damage
-
Other miscellaneous sexual offences
-
Other theft
-
Racially or Religiously aggravated assault with injury
-
Racially or Religiously aggravated assault without injury
-
Racially or Religiously aggravated criminal damage
-
Racially or Religiously aggravated harassment
-
Rape of a female – multiple undefined offenders
-
Rape of a female aged 16 and over
-
Rape of a female child under 13
-
Rape of a female child under 16
-
Rape of a male – multiple undefined offenders
-
Rape of a male aged 16 and over
-
Rape of a male child under 13
-
Rape of a male child under 16
-
Robbery of business property
-
Robbery of personal property
-
Sexual activity etc with a person with a mental disorder
-
Sexual activity involving a child under 13
-
Sexual activity involving a child under 16
-
Sexual assault on a female aged 13 and over
-
Sexual assault on a female child under 13
-
Sexual assault on a male aged 13 and over
-
Sexual assault on a male child under 13
-
Sexual grooming
-
Shoplifting
-
Stalking
-
Theft – making off without payment
-
Theft by an employee
-
Theft from a vehicle
-
Theft from an automatic machine or meter
-
Theft from the person
-
Theft in a dwelling other than from an automatic machine or meter
-
Theft of mail
-
Theft or unauthorised taking of a motor vehicle
-
Theft or unauthorised taking of a pedal cycle
-
Threats to kill
-
Unnatural sexual offences
Positive outcome codes:
The table below are the outcome codes used to calculate the impact of Grip on positive outcome volumes/rates.
Outcome code | Outcome Description |
---|---|
1 | Charge/Summons |
2 | Caution – youths |
3 | Caution – adults |
4 | Taken into consideration (TIC) |
5 | Offender died |
6 | Penalty Notices for Disorder |
7 | Cannabis/Khat Warning |
8 | Community Resolution |
22 | Diversionary, educational or intervention activity |
Annex H: Crime coverage
Year ending 31 March 2022:
Police force area | Hot spot area (km2) | Force area (km2) | % Force covered by hot spots | Violence/robbery offences in hot spots 21/22 | Violence/robbery offences in force 21/22 | % Violence/robbery offences covered by the hot spots |
---|---|---|---|---|---|---|
Avon and Somerset | 0.3 | 4784.3 | 0.01% | 921 | 34,738 | 2.7% |
Bedfordshire | 10.4 | 1235.4 | 0.84% | 3,371 | 12,753 | 26.4% |
Cleveland | 3.0 | 597.6 | 0.51% | 2,846 | 16,484 | 17.3% |
Essex | 3.5 | 3671.5 | 0.09% | 3,480 | 41,303 | 8.4% |
Greater Manchester | 2.0 | 1276.0 | 0.16% | - | - | - |
Humberside | 4.5 | 3515.5 | 0.13% | 3,018 | 22,955 | 13.1% |
Leicestershire | 5.5 | 2550.9 | 0.22% | 2,667 | 23,538 | 11.3% |
Merseyside | 1.3 | 652.2 | 0.21% | 1,454 | 43,527 | 3.3% |
Metropolitan Police | 2.3 | 1570.5 | 0.15% | 4,711 | 181,515 | 2.6% |
Northumbria | 15.6 | 5572.3 | 0.28% | 2,961 | 33,171 | 8.9% |
Nottinghamshire | 46.5 | 2159.3 | 2.15% | 6,021 | 23,946 | 25.1% |
South Yorkshire | 2.3 | 1551.5 | 0.15% | 2,657 | 34,581 | 7.7% |
Sussex | 0.4 | 3786.7 | 0.01% | 820 | 30,698 | 2.7% |
Thames Valley | 2.3 | 5743.4 | 0.04% | 3,108 | 42,155 | 7.4% |
Year ending 31 March 2023:
Police force area | Hot spot area (km2) | Force area (km2) | % Force covered by hot spots | Violence/robbery offences in hot spots 22/23 | Violence/robbery offences in force 22/23 | % Violence/robbery offences covered by the hot spots |
---|---|---|---|---|---|---|
Avon and Somerset | 1.2 | 4784.3 | 0.02% | 2,283 | 33,880 | 6.70% |
Humberside | 3.2 | 1235.4 | 0.26% | 2,293 | 12,382 | 18.50% |
Kent | 2.1 | 597.6 | 0.35% | 2,384 | 18,156 | 13.10% |
Lancashire | 1.2 | 3671.5 | 0.03% | 1,653 | 41,080 | 4.00% |
Leicestershire | 26.6 | 1276 | 2.08% | 8,830 | 85,658 | 10.30% |
Merseyside | 1.6 | 4148.2 | 0.04% | 2,495 | 44,968 | 5.50% |
Metropolitan Police | 4.5 | 3515.5 | 0.13% | 256 | 2,446 | 10.50% |
Northumbria | 50 | 3739.2 | 1.34% | 11,641 | 49,382 | 23.60% |
Nottinghamshire | 17.4 | 3067.1 | 0.57% | 7,263 | 36,889 | 19.70% |
South Yorkshire | 1.3 | 2550.9 | 0.05% | 1,312 | 24,914 | 5.30% |
Sussex | 4.2 | 652.2 | 0.65% | 4,843 | 43,246 | 11.20% |
Thames Valley | 7.8 | 1570.5 | 0.50% | 8,951 | 191,814 | 4.70% |
West Midlands | 61.2 | 5572.3 | 1.10% | 8,411 | 34,286 | 24.50% |
West Yorkshire Police (WYP) | 28.6 | 2159.3 | 1.79% | 5,089 | 24,408 | 20.80% |
References
Andresen MA and Hodgkinson TK (2018) ‘Evaluating the impact of police foot patrol at the micro-geographic level’. Policing: An International Journal: volume 41(3), pages 314-324 [viewed on 20 February 2025]
Barnes GC, Williams S, Sherman LW, Parmar J, House P and Brown SA (2020) ‘Sweet spots of residual deterrence: A randomized crossover experiment in Minimalist police patrol’. SocArXiv [viewed on 20 February 2025]
Basford L, Sims C, Agar I, Harinam V and Strang H (2021) ‘Effects of one-a-day foot patrols on hot spots of serious violence and crime harm: A randomised crossover trial’. Cambridge Journal of Evidence-Based Policing: volume 5(3-4), pages 119-133 [viewed on 20 February 2025]
Bittner E (1970) ‘The functions of the police in modern society’. New York: Aronson
Bland M, Leggetter M, Cestaro D and Sebire J (2021) ‘Fifteen minutes per day keeps the violence away: A crossover randomised controlled trial on the impact of foot patrols on serious violence in large hot spot areas’. Cambridge Journal of Evidence-Based Policing: volume 5(3-4), pages 93-11 [viewed on 20 February 2025]
Braga AA and Weisburd D (2006) ‘Problem-oriented policing: The disconnect between principles and practice’ in Police Innovation: Contrasting Perspectives, pages 133-154. Cambridge University Press [viewed on 20 February 2025]
Braga AA, Papachristos AV and Hureau DM (2010) ‘The concentration and stability of gun violence at micro places in Boston, 1980-2008’. Journal of Quantitative Criminology: volume 26, pages 33-53 [viewed on 20 February 2025]
Braga AA, Papachristos AV and Hureau DM (2012) ‘Hot spots policing effects on crime’. Campbell Systematic Reviews: volume 8(1), pages 1-96 [viewed on 20 February 2025]
Braga AA, Papachristos AV and Hureau DM (2014) ‘The effects of hot spots policing on crime: An updated systematic review and meta analysis’. Justice Quarterly: volume 31, pages 633-663 [viewed on 20 February 2025]
Braga AA, Turchan BS, Papachristos AV and Hureau DM (2019) ‘Hot spots policing and crime reduction: An update of an ongoing systematic review and meta-analysis’. Journal of Experimental Criminology: volume 15(3), pages 289-311 [viewed on 20 February 2025]
Goldkamp J and Vilcica E (2008) ‘Targeted enforcement and adverse system side effects: The generation of fugitives in Philadelphia’. Criminology: volume 46(2), pages 371-409 [viewed on 20 February 2025]
Groff ER, Weisburd D and Yang S-M (2010) ‘Is it important to examine crime trends at a local “micro” level?: A longitudinal analysis of street to street variability in crime trajectories’. Journal of Quantitative Criminology: volume 26, pages 7-32 [viewed on 20 February 2025]
Heeks M, Reed S, Tafsiri M and Prince S (2018) ‘The economic and social costs of crime (second edition)’. Research report 99. London: Home Office [viewed on 20 February 2025]
Hinkle JC, Weisburd D, Famega C and Ready J (2013) ‘The problem is not just sample size: The consequences of low base rates in policing experiments in smaller cities’. Evaluation Review: volume 37(3-4), pages 213-238 [viewed on 20 February 2025]
HM Treasury (2023) ‘GDP deflators at market prices, and money GDP’. [viewed on 20 February 2025]
Home Office (2021) ‘£35.5m to support young people at risk of involvement in serious violence’. [viewed on 20 February 2025]
Home Office (2022) ‘Home Office Annual Data Requirement (ADR) data – Privacy Information Notice’. Home Office Guidance [viewed on 20 February 2025]
Home Office (2024) ‘Evaluation report on Grip and bespoke-funded hot spot policing’ [viewed on 19 February 2025]
Hutt O, Bowers K, Johnson S and Davies T (2018) ‘Data and evidence challenges facing place-based policing’. Policing: An International Journal: volume 41(3), pages 339-351 [viewed on 20 February 2025]
Blanes i Vidal J and Kirchmaier T (2018) ‘The effect of police response time on crime clearance rates’. The Review of Economic Studies: volume 85(2), pages 855-891 [viewed on 20 February 2025]
Koper CS (1995) ‘Just enough police presence: Reducing crime and disorderly behavior by optimizing patrol time in Crime hot spots’. Justice Quarterly: volume 12(4), pages 649-672 [viewed on 20 February 2025]
Koper CS (2014) ‘Assessing the practice of hot spots policing: Survey Results From a National Convenience Sample of Local Police Agencies’. Journal of Contemporary Criminal Justice: volume 30(2), pages 123-146 [viewed on 20 February 2025]
Lum C, Hibdon J, Cave B, Koper CS and Merola L (2011) ‘License plate reader (LPR) police patrols in crime hot spots: An experimental evaluation in two adjacent jurisdictions’. Journal of Experimental Criminology: volume 7(4), pages 321-345 [viewed on 20 February 2025]
Sherman L and Rogan D (1995) ‘Effects of gun seizures on gun violence: ‘Hot spots’ patrol in Kansas City’. Justice Quarterly: volume 12(4), pages 673-694 [viewed on 20 February 2025]
Sherman LW (2022) ‘“Test-as-you-go” for hot spots policing: Continuous Impact Assessment with repeat crossover designs’. Cambridge Journal of Evidence-Based Policing: volume 6(1-2), pages 25-41 [viewed on 20 February 2025]
Sviridoff M, Sadd S, Curtis R and Grinc R (1992) ‘The neighbourhood effects of street-level drug enforcement: Tactical narcotics teams in New York’. New York: Vera Institute of Justice [viewed on 20 February 2025]
Taylor B, Koper CS and Woods DJ (2010) ‘A randomized controlled trial of different policing strategies at hot spots of violent crime’. Journal of Experimental Criminology: volume 7(2), pages 149-181 [viewed on 20 February 2025]
Weisburd D (2015) ‘The law of crime concentration and the criminology of place’. Criminology: volume 53(2), pages 133-157 [viewed on 20 February 2025]
Weisburd D, Telep CW, Hinkle JC and Eck JE (2008) ‘The effects of problem‐oriented policing on crime and disorder’. Campbell Systematic Reviews: volume 4(1), pages 1-87 [viewed on 20 February 2025]
Weisburd D, Telep CW, Vovak H, Zastrow T, Braga AA and Turchan B (2022) ‘Reforming the police through Procedural Justice training: A multicity randomized trial at crime hot spots’. Proceedings of the National Academy of Sciences: volume 119(14), e2118780119 [viewed on 20 February 2025]
Weisburd D, Wilson D B, Gill C, Kuen K and Zastrow T (2024) ‘Are We Underestimating the Crime Prevention Outcomes of Community Policing? The Importance of Crime Reporting Sensitivity Bias’. Journal of Law and Empirical Analysis: volume 1(1), pages 162-179 [viewed on 20 February 2025]
Weisburd D, Wyckoff LA, Ready J, Eck JE, Hinkle JC and Gajewski F (2006) ‘Does crime just move around the corner? A controlled study of spatial displacement and diffusion of crime control benefits’. Criminology: volume 44(3), pages 549-592 [viewed on 20 February 2025]
Wooldridge JM (1999) ‘Distribution-free estimation of some nonlinear panel data models’. Journal of Econometrics: volume 90(1), pages 77-97 [viewed on 20 February 2025]
-
Problem-solving policing uses the SARA (scanning, analysis, response, assessment) model of problem solving. ↩