Testing Application-Based Tasking and Hotspots Policing: A Two-in-One Randomised Trial

Testing Application-Based Tasking and Hotspots Policing: A Two-in-One Randomised Trial | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Testing Application-Based Tasking and Hotspots Policing: A Two-in-One Randomised Trial Tori P. A. Olphin, Owen Miller, Nick Portnell, Lewis Prescott-Mayling, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4650164/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 11 Feb, 2025 Read the published version in Cambridge Journal of Evidence-Based Policing → Version 1 posted 10 You are reading this latest preprint version Abstract Purpose Test the impact of mobile application-targeted patrols of Joint Operations Unit officers at hotspots of violent crime on levels of patrol conducted, and levels of violence in those locations. Methods Crossover-randomized experiment, with hotspots (n = 45) randomly allocated using a mobile phone-based tasking application, with half allocated each day. Impact on levels of patrol and of violent crime were examined using t and chi-squared tests. Results Tasking via app led to dramatically increased officer time in hotspots. An 8.74% decrease in violent crime was seen but was non-significant, and with lower effect sizes than have been found elsewhere. Conclusion Using a tasking application provided a cost-effective mechanism for achieving hotspots patrols using business-as-usual resources. Traditionally designed hotspots didn’t appear optimal for policing violent crime in Thames Valley, and these need to be redesigned to work for the geography and crime density in Thames Valley. Further research should test redesigned hotspots. Hotspots Policing Prevention of Crime Crime Reduction Randomised Controlled Trial Violent Crime Application-based Tasking Figures Figure 1 Figure 2 Introduction This study is concerned with randomised testing of hotspots policing, using targeted patrols of Joint Operations Unit officers to hotspots of violent crime through use of a newly developed mobile phone application in Thames Valley Police, the largest non-metropolitan policing agency in the United Kingdom. Violent crime has been recognised for a long while by the World Health Organisation ( 1997 ) as a serious international health concern, with clear connections to disability, personal injury and illness. The Office for National Statistics ( 2023 ) shows that there were an estimated 0.9 million violent crimes in the year ending June 2023, and that this was no significant change from the previous year, indicating a massive amount of harm that is being done to members of the UK public each year due to interpersonal violence. There are a number of mechanisms or theories that have led to the development of hotspots policing as an approach to reduce crime, and that we can take into account alongside the existing evidence to build a strong methodology with which to test hotspots policing in the Thames Valley. These mechanisms are that people can be deterred from committing crimes, and that crime is concentrated in some places more than others. Deterrence theory suggests that potential offenders can be prevented from committing crimes when the perceived costs of committing the offence outweigh the perceived benefits from doing so (Zimring and Hawkins, 1973 ). This relates to rational choice theory (Pratt, 2008 ), which suggests that people think through their offending, and make decisions about whether to commit criminal offences based on the likelihood of being caught or punished. Routine activities activity (Cohen and Felson, 1979 ) suggests that for crime to take place, there needs to be a victim and an offender in the same place at the same time, and this needs to occur in the absence of a capable guardian. Nagin and colleagues ( 2015 ) argue that increasing visibility of police officers in hotspots will create deterrent effects by increasing the perceived likelihood of apprehension. Based on all of the above, for hotspots policing to be effective at reducing violent crime, there would be a requirement for the offender to either notice a patrolling officer or be uncertain as to the lack of presence of an officer, and feel that the risk of committing such an offence would outweigh the benefit or lack of thought taken to commit it. It is likely that hotspots policing may be more difficult, and potentially less effective, for prevention of violent crime due to the prevalence of crimes involving alcohol or drugs (Duke et al., 2018 ), and findings that violence is more likely to occur when alcohol has been consumed than when it has not been (Haggård-Grann et al., 2006 ). Alcohol or drugs may lead to decreased inhibition (Duica et al., 2020 ), and in turn may reduce the deterrence effect or at least a lingering deterrence effect from having seen an officer earlier in the evening. A second factor that may make non-pre-meditated violence more difficult to deter is where violence occurs due to a personal insult and immediate violent response (Cohen et al., 1996 ). This fast thinking or not thinking reaction may well be less easy to deter due to a lack of consideration of consequences (Kahneman, 2011 ). Additionally, many violent offences occur in buildings such as pubs or clubs, and it may be more difficult to provide a deterrent effect within buildings. This is consistent with the findings of Braga and colleagues ( 2019 ) which showed that mean effect size of hotspots policing on violent crime (0.102) is lower than the mean effect size on drug (0.244), disorder (0.161) or property-related (0.124) crime outcomes. Consistent with Rossi’s ( 1987 ) Stainless Steel Law of Evaluation which suggests that the better designed the impact assessment, the more likely it is to find net impact to be zero, they also found that randomised controlled trials were associated with lower mean effect sizes (0.109) than quasi-experimental designs (0.171). The consistent finding that both crime (Sherman et al., 1989 ) and crime harm (Weinborn et al., 2017 ) are concentrated in some areas more than others has led Weisburd ( 2015 ) to suggest that there is a law of crime concentration at places, and that this occurs to a similar degree across different cities and throughout time. This remains true even in areas that have extremely high levels of crime, with the majority of the space being relatively crime free and a small percentage of the area accounting for the vast majority of the crime (Weisburd et al., 2012 ). This, added to the theory that capable guardians being present at a location can deter potential offenders from committing crime provides a very compelling argument for hotspots policing, as succeeding at deterring crime at locations which account for a high proportion of all crime would have a large impact on the overall level of crime in a city whilst deploying officers efficiently. There is a significant quantity of evidence (Braga et al., 2019 ) from hotspots policing trials in the United States which suggests that especially for drug crime, disorder and property crime, hotspots policing is associated with decreased levels of crime. Despite having an extensive evidence base from the United States, we still have significant gaps in our knowledge of what works in the United Kingdom. Of the 65 eligible hotspots policing evaluations included in meta-analysis by Braga and Colleagues ( 2019 ), 51 were conducted in the US, and only 4 in the UK (Ariel et al., 2016 ; Ariel and Partridge, 2017 ; Fielding and Jones, 2012 ; Williams, 2015 ). Since the cutoff date for that meta-analysis, these authors were able to find two more randomised trials (Basford et al., 2021 ; Bland et al., 2021 ), one more publication of one of the trials in the meta-analysis (Williams and Coupe, 2017 ), and two more quasi-experimental trials (de Brito and Ariel, 2017 ; Gibson et al., 2017 ) of hotspots policing that that have been conducted in the UK and published, so the evidence base is growing, but there are still significant evidence gaps in relation to the efficacy of hotspots. There were some interesting findings from these UK based studies that have also gone into our thinking for design of this research. Basford and colleagues found that in small (150m 2 ) hotspots of assault, robbery and drug dealing, it was possible to reduce crime harm on days with patrols by operational support group officers, generally experienced officers who are well practiced in proactive policing. Williams and Coupe ( 2017 ) found that even when total patrol time was controlled for, fewer longer (mean 2.5 patrols of 9.6 minutes) patrols were associated with lower levels of crime than hotspots with more shorter patrols (mean 5 visits averaging 5.2 minutes each). This adds further weight to Koper’s ( 1995 ) correlational findings that patrols of 10–15 minutes appeared optimal for hotspots patrols to reduce crime. Ariel and colleagues ( 2016 ) found that additional patrols by police community support officers, a visible patrol officer with limited powers of arrest, also led to decreased crime, suggesting that the visibility of a capable guardian may be more of a factor than the powers at the disposal of that capable guardian. Ariel and Partridge(2017) showed potential for a backfire effect in relation to hotspots policing and hypothesised that this may be due to identifiable patrol patterns. However, based on the level of power found in other hotspots trials, the lack of a crossover design in this trial may have meant that it was simply underpowered and whilst we will remain aware of the potential for a backfire effect, these findings are not of major concern in design of this trial. Unfortunately, whilst some of the other studies tested interesting concepts such as reducing patrol(Gibson et al., 2017 ) and the effect of feedback and tracking on delivery of patrols (de Brito and Ariel, 2017 ), both of these trials were quite underpowered, so it would not be possible to draw generalisable conclusions from them. Whilst there have been strong effects reported by Bland and colleagues ( 2021 ), who found that it was possible to prevent serious violence across a large area with minimal amounts of foot patrol, this study was conducted during periods of lockdown in the UK. The presence of lockdown restrictions is likely to have altered both the underlying level of crime due to reductions in the numbers of people present in public, and the general presence of police officers on routine patrol. In addition, when people were out in public, they had to be outdoors to congregate which will make it much easier for officers to be seen. Therefore it is possible that these findings were an anomaly due to it being easier to see police officers, easier for police officers to see incidents occurring, and a lower baseline level of patrol at the time. It is therefore suggested that the best existing evidence would be for around three patrols per day each lasting approximately 15 minutes, in small hotspots where the officer can be seen by anyone in the hotspot for at least part of the patrol if not the majority of it and that this would be a good place to start testing, given that we do not currently know what works best for hotspots policing in non-metropolitan areas of the United Kingdom, especially in policing areas with a massive land area. Despite the high level of evidence surrounding hotspots policing, in the UK there are still many unanswered questions and, in areas that have implemented hotspots policing, there are often large additional costs such as GPS trackers which may prohibit large scale implementation. Consistent with other researchers (Basford et al., 2021 ; Bland et al., 2021 ), at the start of this trial we were not aware of any UK police agency that had a force-wide patrol strategy to target hotspots of violence that they were tracking delivery of on a daily basis. We were also not aware of any test involving implementation of place-based policing using a mobile phone application, that linked securely into police systems, as the tasking mechanism; something that allows for much more cost-effective implementation, and which would remove the technological barriers to implementation of hotspots policing raised by Ariel ( 2023 ), and allow for tracking of delivery, argued to be essential in any implementation (Sherman, 2013 ), to be automated. Therefore, it is the aim of this research to design an implementation that is the best possible test of the existing best-evidence in a way that can be rolled out to business-as-usual across the entire force area for low to no additional cost through concurrent testing of a new hotspots policing mobile phone app. Methods Thames Valley Police is the largest non-metropolitan police force in the United Kingdom, covering over 2,200 square miles and three counties; Berkshire, Buckinghamshire and Oxfordshire. This massive area and range of urban and rural areas meant that careful thought needed to go into how additional policing could best be delivered to hotspots spread across the Thames Valley, especially given the large distances between hotspots that would make it incredibly difficult to have dedicated hotspots patrol resources moving between hotspots. Choice of Operational Resource Limiting the experiment to individual parts of the force area would not have elicited sufficient locations to provide analytic power capable of showing a result. To deliver consistent policing of hotspots across the entirety of the Thames Valley, using officers who were equipped to travel to the hotspots and whose patrols could be managed to avoid treatment of hotspots on days where those hotspots were in the control group, it was decided that the Joint Operations Unit (JOU) would be the most appropriate resource for conducting patrols. The JOU is a shared resource with Hampshire Constabulary, and comprises roads policing unit, firearms capability, operational support, canine unit, and mounted unit. The JOU is a flexible resource, used to conducting disparate types of policing operation, and they are well equipped to travel across the Thames Valley to conduct policing activities. In addition, it was seen as essential that we were able to manage the resources effectively and ensure that all areas were receiving comparable policing of hotspots. The JOU allowed us to use one command structure to achieve this, avoiding large amounts of liaison between different command structures as would be the case if conducting patrols across different policing areas using their own local resources. Patrolling was completed during both downtime within the JOU’s core shift time (i.e. time between attending incidents), and in periods of funded overtime offered to JOU officers to focus specifically on the violent crime hotspots. Selection of hotspots The randomised controlled trial was planned for implementation during the period from September 2021 to March 2022, and to avoid introducing seasonal hotspots that would not be relevant to the time period of the RCT, the data used to build hotspots was restricted to occurrences that took place only during these months from September 2016 to March 2020. Data from 2020 to 2021 was not used, due to the unknown effects of lockdowns during the covid-19 pandemic. A visual search of the locations of incidents was conducted, and the locations of serious violent occurrences was consistent with that of violent offences with less serious outcomes. It was decided that all violent acts, from public order offences all the way up to murder, would be incorporated into the data used to build hotspots, so that locations would not be skewed by small numbers of incidents which would not be preventable during the test period. Domestic violence offences were removed from the dataset, as these were deemed be much less preventable through patrol, due to high prevalence of these crimes occurring inside private dwellings. An initial dataset of incidents (n = 59,103) was examined for spatial quality in terms of whether the location was able to be plotted on a map using either XY coordinates or postcode. Where the location was able to be plotted using XY coordinates or postcode, these incidents were geocoded and mapped (n = 58,959). To control for likely differences in violent crimes that occur in the daytime and in the evenings, the data were split into two periods; days (from 08:00 to 19:59, n = 35,494) and nights (from 20:00 to 07:59, n = 23,465). These were the final datasets for identification of hotspots. Once hotspots were identified, it became apparent that all the days’ and nights’ hotspots were different areas, and this supported the decision to split the data and identify spatio-temporal hotspots. To allow for us to test whether a ‘standard’ size of hotspot (one where an officer can be seen in all of the hotspot during a patrol) is an effective mechanism for hotspots policing in the Thames Valley, a tessellation of hexagons was created for the entire Thames Valley Police area. Crime levels in these hexagons were examined to ensure that the size of hotspot could be kept as small as possible. This would ensure that the entire hotspot could be patrolled during a fifteen-minute patrol, whilst still containing sufficient crime that a reduction could realistically be detected through testing. This reached a balance where hexagons with edge length of 150m were selected as being appropriate in both measures. This ensured that sufficient crimes were included in the highest concentration hotspots, and that they could be patrolled in a manner that officers could be seen from the majority of the hotspot while they were patrolling, maximising the impact of patrols. This resulted in just under 145,000 hexagons across the force area. Incidents were aggregated into the hexagons, and the top 50 hotspots based on crime count for days and nights were retained for manual examination. Hexagons which included hospitals, prisons and schools were removed manually, as the majority of crimes in these hotspots are likely to have occurred indoors, and would not be preventable through patrols. A seventy-five metre buffer zone was applied to each hotspot, so that cross-contamination of patrols could be minimised. Sanity checks were conducted to ensure that these hotspots were enduring (consistent over time), that they remained consistent when examined using different methods of hotspot identification (clustering and optimised hot spot analysis), and that they did not miss high concentrations of crime harm. Mapping against Cambridge crime harm index scores (Sherman et al., 2016 ), and year by year examination did not change the locations of the selected hexagons, and there were no additional areas that appeared to be as appropriate as the ones selected. This was kept in sight whilst the remaining hotspots were reviewed and manually manipulated (rotated and/or nudged spatially) to ensure that; there were no overlapping buffer areas, areas such as dual carriageways which could not be patrolled were removed, and the incident inclusion was maximised. This spatial nudging was done to ensure that the hotspots were in the most appropriate places, not just where the tessellation had originally placed them, and was done with view of all crime incidents on the map at the same time. Aggregated totals were recalculated, and the research team reviewed all the final hexagon locations with tactical leads to ensure that the hotspots made sense from their point of view. Forty-five hotspots were identified for the experiment (days n = 19; nights n = 26). This was accepted by tactical decision makers as being reasonable for delivery of patrols. Power analysis calculations were conducted to establish the likely case count that would be required to show an effect using previously found effect size of 0.102 for impact of hotspots policing on violent crime (Braga et al., 2019 ), and likely standard deviation of control hexagons of 0.01. According to these calculations, the expected number of cases required in each group in order to show an effect as calculated above, at the mean positive effect size of previous studies is 2021 for power of 0.9, and 1510 for power of 0.8. That means that between 3020 and 4042 total treatment or control periods would be required. Tasking, targeting and tracking via a mobile phone application Having identified a roaming resource to carry out the hotspots patrols as additional policing and the hotspots that they would be tasked to patrol, it was necessary to design a mechanism by which officers could be briefed on the areas to visit, and hotspots were randomised and correctly displayed to officers. There was also a necessity to record how many patrols were conducted to track the patrols. This research team is aware that in other areas of the UK, control room resources are used to direct officers into the hotspots, and this was considered as an option. However, the view of this research team is that this should not be a test of what is possible if we throw money at a problem, but rather what solution might be possible that would deliver a long-lasting effect, and that could be made into a business-as-usual delivery, with no additional cost. This would mean that it can be scaled up to deliver a lasting improvement in service and efficiency, rather than something that would disappear if funding were no longer available. To satisfy the requirements of targeting the appropriate hotspots, tracking the visits when they occur, and creating a tasking mechanism that is sustainable and scalable, members of this research team designed a hotspots patrol application using Microsoft Power Apps. This application was logged into by officers on their force mobile phones, and presented them the hotspots that were allocated for patrol that day or night. It showed them an overview map to help them get to the location, and a detailed map of what was in the hotspot, as well as a postcode for satellite navigation and details of any warning markers to look out for. Figures 1 and 2 below shows two screenshots from the live version of the app, as it would be seen by officers. When officers started their patrol in a hotspot, they click a button to start the recording of the patrol, and a timer popped up to show them how long they had been actively patrolling. This turned amber when it got to 13 minutes, then green at 15 minutes. This was designed to add an element of a nudge effect (Thaler and Sunstein, 2009 ) to work in addition to training about what length a patrol should be, and encourage patrols of a length that would match that suggested by Koper ( 1995 ) as being the optimal length to see a deterrent effect that lasts. When the patrol finished, they clicked again, and then used sliders to provide details of patrol activities carried out (positive engagement, stop and search, and arrests). This is therefore different from most other hotspots policing interventions which are quite resource intensive for management of the intervention. This experiment is testing both whether tasking officers using an application on their mobile devices results in additional patrols, and whether this in turn creates a change in the amount of violent crime in small hexagonal hotspots within Thames Valley Police when roaming joint operations unit officers are tasked to patrol them. The data from the application was fed through a Microsoft Azure environment to a Power BI dashboard where it was updated every two hours so our research team could track delivery of treatment during the experiment. This dashboard was reviewed with our tactical leads twice a week to ensure that treatment fidelity was as high as possible. A backup mechanism using Microsoft Forms was also created in case of failure of the app, though this was fortunately not needed during the delivery of the trial. In addition, archive data from airwave radios was collected for the hotspot areas so that the level of overall patrol in the hotspots could be examined, along with the ratio of pings from JOU radios between the test and control hotspots. Testing: how the experiment was conducted The randomisation was a cross-randomised design, where hotspots were randomised each day to one of treatment or control, and each hotspot will therefore appear as treatment and control dependent on the day or night in question. Cases were pre-randomised using the sample() function in R (Ihaka and Gentleman, 1996 ). Pre-randomisation was used so that the randomisation could be incorporated into the app and the dashboards. However, the actual randomisation was blind to both researchers and officers until the point that the application displayed the treatment hotspots on the day. The randomisation was a simple randomisation each day to ensure that the number of treatment hotspots are consistent (9/10 out of 19 for days, and 13 out of 26 for nights), This would keep the experiment within bounds that were agreed with Joint Operations Unit in relation to the number of hotspots they believed they would be able to effectively patrol. Officers were given guidance at the beginning of the experiment in how to use the app, and what was expected of them. They were briefed that repeated patrols (3 per hotspot per day) were desirable, that patrols should last between 13 and 20 minutes, and that they should be concentrating their patrols on the hotspots that were allocated through the application, not based on previous days. The test period was pre-defined as being between 08/11/2021 and 31/03/2022 inclusive. This time period was a constraint of the experiment, due to funding windows. It was decided at the start of the experiment that the outcomes should be blind to the research team until the experiment was concluded, in order that knowledge about how crime was being affected could not be used in any way to alter the integrity of the trial. Three factors arose at one point during the RCT which would have severely limited the ability to provide resources for the experiment between 19/12/2021 and 03/01/2021 inclusive; Omicron Covid-19 variant limited overall staffing levels, overtime was paused meaning no additional staffing, and there were some issues with vehicles (Roberts, 2023 ) which limited patrols by JOU units. Therefore, it was decided that the period between 19/12/2021 and 03/01/2021 inclusive would not be part of the RCT, and no measurement would be taken from this period. No analysis of crime levels was done prior to making this decision, so the researchers were blind to the ongoing performance of the trial, and the decision was made in advance of the problems occurring due to identification of the potential issues during our bi-weekly planning meetings. It was therefore believed to be fair and reasonable to make this decision at the time. Analysis plan This experiment is examining two different outcomes; whether the tasking of officers via a mobile phone application results in an increase in patrol time, and in turn whether this then has a knock-on effect on crime. To answer the first part, data from the application will be used to show the mean number of patrols per day and give an overview of the level of patrol that was added through development of the application. In addition to this, airwave pings will be examined when they occur within the hotspots. The airwave pings also record the speed of travel at the time, and since we are looking for patrols that have the potential to deter criminal activity it was decided that pings over 50km/h should not be counted. Therefore, the mean pings in the hotspots per day will be examined both for joint operations unit officers, and for all other officers to provide a baseline of the activity in those areas. To answer the question of whether this approach has then led to a change in the level of crime, three different types of crime will be examined. Firstly, violent crime is defined for this research as any offence against the person where violence is used, or feared that it will be used. Therefore, this includes all offences from S.4 Public Order Act, up to and including Homicide. The second category adds in sexual offences to the above group, as the level of violence used in a common assault and a sexual assault may well be the same, yet sexual assaults are not included in most examinations of violent crime. This may therefore reduce the visibility of women and girls as victims of public place offending. To avoid this, all offences of sexual assault, with or without penetration, and rape will be included in a violent and sexual offences category. Thirdly, total recorded crime, including all offences, will be examined to establish whether there is an overall impact over and above that seen in terms of violent crime. All three of the above crime types will be examined in terms of their prevalence, or whether there is one or more crime of that category in that location on that day, the mean count per day of that category, and the mean harm as measured by the Cambridge crime harm index (Sherman et al., 2016 ). All of the above analyses of both change in patrol and change in crime will be conducted as an overall combination of all days and nights hotspots, and will also then be broken down to examine days and nights independently to establish whether there are any differential levels of impact at different times of day. The analyses are based on a pure intent to treat approach, and therefore the analysis will be conducted based on the randomised allocation, regardless of level of patrol that was conducted. The intent to treat approach is used because this is a test of an implementation of an intervention in a live environment, and it is therefore not only important to establish whether it works when implemented appropriately, but also how it works in the real world (Detry and Lewis, 2014 ), as the implementation of a low-contact tasking mechanism for officers via a mobile phone application is what is being tested, and it would not be a fair test if analysis were restricted only to times where it was used correctly. Findings In this research, two main questions were asked; firstly, does tasking officers to attend hotspots of violent crime, using a mobile phone application, result in more patrols? Secondly, when hotspots of violence were identified using best practice found in previous research, and officers are patrolling them, does this result in a reduction in violent crime? As can be seen from Table 1 , there was an average of 1.255 patrols of at least 13 minutes per treatment hotspot per day by officers using the application, totalling over 3,500 additional patrols. To assess whether there was a change in actual time in the hotspots, the recorded location pings from airwave radios were compared, and there were significantly more airwave pings by joint operations unit officers in the treatment hotspots than in the control hotspots, t(5162.8) = 18.76, p < 0.05. This was an increase of 93% over the control hotspots, increasing from 5.89 pings per control hotspot per day to 11.39 pings per treatment hotspot per day. Joint Operations Unit officers only account for a small proportion (around 10%) of frontline officers in Thames Valley Police, so whilst this did also increase the overall amount of policing activity when all officers were included, this increase was non-significant. Whilst the above differences in patrols and airwave pings were seen in both days (Table 2 ) and nights (Table 3 ) hotspots, the increase in patrol was greater for days hotspots than nights hotspots. There were 2.01 patrols on average per treatment hotspot per day for days hotspots, but only 0.713 for nights hotspots. The mean count of airwave pings increased from 4.06 to 13.13 pings per hotspot per day for days hotspots, t(1835.9) = 21.864, p < 0.05, and increased from 7.24 to 10.11 pings per hotspot per day for nights hotspots, t(3124.4) = 7.168, p < 0.05. Additionally, it was found that the vast majority of patrols exceeded the desired 15 minutes as demarcated by the change in colour of the timer in the app. Figure 3 shows the distribution of patrols of different lengths, which can clearly be seen to be a non-normal distribution. In relation to the levels of crime seen in the treatment and control hotspots, Table 4 shows that whilst there appears to have been a decrease in the prevalence of crime in the hotspots, this was not a significant finding. For the sake of transparency, all analyses are shown below. From this, it can be seen that there was on average only one crime of any type every two days, and this decreased for violent crime to one every 5 days in the areas that were being patrolled. Table 5 and Table 6 show that these findings were consistent across days and nights, with violent crime being more prevalent during the nights hotspots overall, but it appears that despite the hotspots being created in a manner consistent with prior research, and these being the areas of highest concentration of violent crime, the overall prevalence of crime in Thames Valley hotspots created in this way was still low. These findings offer many options for development of this approach, and indicate that it is possible to develop an extremely cost-effective tasking mechanism for targeting of geographical locations. Discussion This randomised trial of the implementation of hotspots policing using a mobile phone app to direct police activity has demonstrated that it is possible to remove many of the barriers to implementation of hotspots policing using a well-designed application that is created with use by officers in mind. The application allowed for officers to be tasked and tracked, on a business-as-usual basis, in a way that did not require additional resourcing that wasn’t delivering the patrol. An alternative would be to use supervisory or control room staffing to direct the patrols in real time. This can become a cost-burden for police services, without adding patrol value. It is therefore believed that this method shows a good compromise between in-the-moment resource allocation and tracking. However, it is acknowledged that there can be other benefits of using additional tasking resources; they would allow for tracking feedback to be given in the moment of the patrol if needed, which could improve quality of patrols. The regular tracking of patrols was conducted using airwave radio data, and this provided a balance between low cost and no resource requirement, and up to date feedback for officers. It is also possible that the application could be developed further to allow for tasking of other types of patrol, for problem-oriented policing activity and tracking of actions taken, as well as for other non-immediate taskings of which police officers have many, so it is a scalable solution, with little-to-no additional cost requirement. There was an error caused by application data not refreshing if the app was not closed overnight which led to a small number of patrols of hotspots on control days. This is likely to have been a limitation in the mechanism used for access of the application, as they were not available to be installed on phones, and had to be accessed via a web application. This should be fixed for future versions to ensure that there is no bleed of treatment into control areas. The application has been released for use by other UK police forces, and is also being developed further to ensure that this tasking mechanism can be used for operational deployments across Thames Valley Police. Both of these will ensure that the aim of the project which was to develop and deliver sustainable mechanisms for tasking of officers to hotspots of crime is successful. This is a technology that is easily scalable, and can be applied in many different aspects of policing. The nudge effect (Thaler and Sunstein, 2009 ) of having the timer change colour at 15 minutes appears to have been extremely beneficial, but needs specific additional research. It also lends credence to using this type of approach in other areas of delivery in the public sector. However, there is a need to be incredibly careful to avoid backfire effects often associated with gaming of things that are seen as targets (Grace, 2022 ). Whilst overall, the use of an app appears to be an effective mechanism for tasking of police resources, there was a larger increase in airwave pings, which can loosely be interpreted as police presence, on days than on nights, indicating that there may be a limitation on resource availability when the app is used later in the day. This is likely to have been compounded slightly due to the timings of overtime shifts which were offered between 2pm and 10pm each day. However, this was a substantial increase in directed policing by resources that would not have been in these areas otherwise (both through business-as-usual resourcing patrolling during downtime, and in directing overtime resources to specific tasks or locations). It is therefore likely that this approach can be expanded out to all business-as-usual resources to allow improved tasking at a whole-force level. It may be possible to improve the tasking further, if the application can be developed to show current location in relation to the hotspot. This may act as a further nudge to ensure that officers keep patrolling the hotspot while they are allocated to. There was an overestimation of how much patrol could reliably be provided, especially on night shifts, and this did impact upon what was possible. It would be beneficial for all forces to establish a baseline of activity for officers who are to be used for hotspots policing, prior to launch of hotspots policing strategies. It also appears that, whilst non-significant, this approach of using joint operations unit officers to patrol small hotspots has led to a reduction in prevalence of total crime and violent crime. This benefit was minimal however, likely due to the low rate of crime commission in those hotspots, as low base rates of crime present significant challenges to evaluation of hotspots policing (Hinkle et al., 2013 ). It was useful to commence testing of hotspots policing using small micro-hotspots, as a test the historic hotspots literature for an area such as Thames Valley Police. However, whilst these hotspots did allow for the entire area to be patrolled by a single officer within 15 minutes, there is a compelling argument for testing of larger hotspots which would need to be contiguous areas that were straightforward to patrol. The expansion of hotspot size would then allow for the quantity and rate of crime in each hotspot to be sufficient that it would be reasonable to expect one crime per day in a hotspot, meaning that there are no days where patrol was conducted but there would not have been a crime in any case. In this study, the rate of crime was much lower than this, and this likely contributes to the lack of significant decreases in crime. The maximum size of these hotspots still needs to allow for officers to be seen across the entire area, as if officers are not seen, they will not create a deterrence effect. It may also be valuable to assess the impact of patrols which enter shops and licensed premises, to increase visibility. This, along with assessment of the quality of patrol, level of visibility of officers, and proactivity displayed might also play a part in the effectiveness or lack thereof in studies of this type. This was not possible for us to assess, though this would be beneficial where possible in future work. When creating hotspots, the level of crime present in the hotspots needs to be preventable by an occasional resource, at the same time as there needs to be a way in which the patrol or action taken could realistically lead to impact on the crime that is desired to be prevented; if the patrol is not visible, and actively patrolling, it is unlikely to impact on the commission of crime. There are limitations in relation to hotspot creation, as it is not currently possible to identify which crimes have occurred indoors and outdoors. This makes it more difficult to base the hotspots solely on crimes that are likely to be preventable, and it would be of benefit for policing agencies to be able to differentiate between crimes committed indoors and outdoors, and in public and private places. An additional way to improve the development of hotspots would be to incorporate ambulance records for incidents without a corresponding crime incident into the data used to build the hotspots. Sutherland and colleagues ( 2021 ) found that approximately 90% of ambulance records for violent crime-related injury attendances did not have a corresponding police record, and that this was even lower for emergency department records. Their findings suggested that adding medical data to police data could increase levels of recorded violence by as much as 15 to 20%, and this in turn could aid the building of hotspots. A further analysis was conducted in the Thames Valley (Simmonds et al., 2023 ), which found that approximately 80% of ambulance records for violence did not have a corresponding police record within 100m of the ambulance attendance site. Whilst the findings of this research are incredibly promising in relation to crime prevention, and have demonstrated that use of an application is a sound method for tasking patrols to hotspots, the hotspot size and crime density was not optimal. The use of experienced proactive officers is suspected to have assisted in the success of this trial, so it is necessary to test how this method would work at a whole-force level, with all officers being able to access the app and patrol larger hotspots ensuring higher base rates of crime. These changes are being incorporated into a second randomised trial which will be rolled out across the Thames Valley, allowing for an approach to be built where results can be learned from and acted upon. Conclusions This research has demonstrated that it is possible to direct police patrols using a mobile phone-based application, and that this produces an incredibly cost-effective mechanism for tasking of police resources, particularly when directing them to specific geographic areas. There are many benefits, that have already been realised in Thames Valley Police, from this; firstly, it is possible to direct officers reliably to areas where additional policing is needed, without additional cost for control room staff, and without additional impact on supervisory staff. It is also possible to use airwave data, examined in an automated manner in the cloud, to track the delivery of tasked activity and as demonstrated here, to track experimental delivery. By using a mechanism such as this app, it is also possible to increase patrol by existing resources, increasing delivery of policing activity using business as usual resources without increasing cost. This was a key driver in the development of this approach, as it is these researchers’ opinion that we need to drive towards embedding best practice within existing mechanisms of policing, ideally designing research in a way that it can be delivered without additional cost and is realistically designed to allow implementation exactly as tested. It is the lead researcher’s strong belief that it is better to deliver the best test of something that can be implemented, than a perfect scientific trial of something that only works in laboratory conditions. However, it has also shown that the optimal mechanism for hotspots policing in violent crime hotspots differs somewhat between different types of policing area, showing a dramatically smaller effect size for treatment of hotspots in a non-metropolitan police force than might have been expected. It is hypothesised that a different style of hotspots, with larger contiguous areas containing a slightly lower concentration but larger overall frequency and count of violent crime would be more effective in this type of police area. This approach is being developed for further testing in a follow-up experiment designed to examine the optimal mechanisms for delivery of hotspots policing in non-metropolitan areas with lower frequency of crime than in large cities. The application has been developed further and as well as being expanded out for use by all officers in this force, it has also been made available to other police forces in the UK so that they may also benefit from a revolutionary way of tasking policing resources to geographical areas. This will allow for a dramatic reduction in cost, and increase in reliability for delivery of hotspots policing approaches. Declarations Funding and Support : This would not have been possible without funding and support from Home Office GRIP Programme, and the strategic and operational support from Thames Valley Police. Any findings, opinions, conclusions, or recommendations expressed herein represent those of the authors and do not necessarily reflect the views of Thames Valley Police, or UK Home Office. Author Contribution T.O was the lead researcher and oversaw design, tracking and analysis of the experiment, and wrote the main manuscript text, figures and tables. O.M and L.P-M designed the experiment with T.O, and ensured delivery was maintained across Thames Valley Police. O.M wrote the application that was used for tasking, and all authors tested the application, both for technical delivery and geographical accuracy. N.P. is the geographer who conducted analysis of crime data and determined the location and spatio-temporal definition of the hotspots. C.Y and J.H led delivery of the patrol-related project and were responsible for liaison with command teams and local policing areas. All authors reviewed the manuscript. Acknowledgement This would not have been possible without funding and support from Home Office GRIP Programme, and the strategic and operational support from Thames Valley Police. We are grateful to the officers from the Joint Operations Unit at Thames Valley Police and Hampshire & Isle of Wight Constabulary who conducted these patrols. Any findings, opinions, conclusions, or recommendations expressed herein represent those of the authors and do not necessarily reflect the views of Thames Valley Police, or UK Home Office. Data Availability The data that support the findings of this study are available from Thames Valley Police, but restrictions apply to the availability of these data, which are derived from sensitive operational data and so are not publicly available. Summary outputs of the data used, and the code used to analyse these, is available from the authors upon reasonable request and with the permission of Thames Valley Police. Norwegian Social Research (NOVA), but restrictions apply to the availability of these data, which were used under licence for the current study and so are not publicly available. The data are, however, available from the authors upon reasonable request and with the permission of Norwegian Social Research (NOVA). References Ariel, B. 2023. Implementation issues with hot spot policing, International Journal of Law, Crime and Justice , vol. 75, 100629 Ariel, B. and Partridge, H. 2017. Predictable Policing: Measuring the Crime Control Benefits of Hotspots Policing at Bus Stops, Journal of Quantitative Criminology , vol. 33, no. 4, 809–33 Ariel, B., Weinborn, C., and Sherman, L. W. 2016. “Soft” policing at hot spots—do police community support officers work? A randomized controlled trial, Journal of Experimental Criminology , vol. 12, no. 3, 277–317 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 , vol. 5, nos. 3–4, 119–33 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 , vol. 5, nos. 3–4, 93–118 Braga, A. A., Turchan, B., Papachristos, A. V., and Hureau, D. M. 2019. Hot spots policing of small geographic areas effects on crime, Campbell Systematic Reviews , vol. 15, no. 3 Brito, C. de and Ariel, B. 2017. Does Tracking and Feedback Boost Patrol Time in Hot Spots? Two Tests, Cambridge Journal of Evidence-Based Policing , vol. 1, no. 4, 244–62 Cohen, L. E. and Felson, M. 1979. Social Change and Crime Rate Trends: A Routine Activity Approach, American Sociological Review , vol. 44, no. 4, 588–608 Cohen, D., Nisbett, R. E., Bowdle, B. F., and Schwarz, N. 1996. Insult, Aggression, and the Southern Culture of Honor: An “Experimental Ethnography,” Journal of Personality and Social Psychology , vol. 70, no. 5, 945–60 Detry, M. A. and Lewis, R. J. 2014. The Intention-to-Treat Principle: How to Assess the True Effect of Choosing a Medical Treatment, JAMA , vol. 312, no. 1, 85–86 Duica, L., Dragulescu, V., and Pirlog, M. 2020. NEUROBIOLOGICAL ELEMENTS OF HOPELESSNESS AND IMPULSIVITY IN SUICIDAL BEHAVIOR, International Journal of Medical Reviews and Case Reports , vol. 4, no. Reports in Clinical Medicine and, 1 Duke, A. A., Smith, K. M. Z., Oberleitner, L. M. S., Westphal, A., and McKee, S. A. 2018. Alcohol, Drugs, and Violence: A Meta-Meta-Analysis, Psychology of Violence , vol. 8, no. 2, 238–49 Fielding, M. and Jones, V. 2012. ‘Disrupting the optimal forager’: predictive risk mapping and domestic burglary reduction in Trafford, Greater Manchester, International Journal of Police Science & Management , vol. 14, no. 1, 30–41 Gibson, C., Slothower, M., and Sherman, L. W. 2017. Sweet Spots for Hot Spots? A Cost-Effectiveness Comparison of Two Patrol Strategies, Cambridge Journal of Evidence-Based Policing , vol. 1, no. 4, 225–43 Grace, S. 2022. The perverse impact of performance measures on policing: lessons from the rise and fall of out of court disposals, Policing and Society , vol. 32, no. 2, 200–220 Haggård‐Grann, U., Hallqvist, J., Långström, N., and Möller, J. 2006. The role of alcohol and drugs in triggering criminal violence: a case‐crossover study*, Addiction , vol. 101, no. 1, 100–108 Hinkle, J. C., Weisburd, D., Famega, C., and Ready, J. 2013. The Problem Is Not Just Sample Size, Evaluation Review , vol. 37, nos. 3–4, 213–38 Ihaka, R. and Gentleman, R. 1996. R: A Language for Data Analysis and Graphics, Journal of Computational and Graphical Statistics , vol. 5, 299–314 Kahneman, D. 2011. Thinking, Fast and Slow , New York, Farrar, Straus and Giroux Koper, C. S. 1995. Just enough police presence: reducing crime and disorderly behavior by optimizing patrol time in crime hot spots, Justice Quarterly , vol. 12, no. 4, 649–72 Nagin, D. S., Solow, R. M., and Lum, C. 2015. Deterrence, Criminal Opportunities, and Police, Criminology , vol. 53, no. 1, 74–100 Organisation, W. H. 1997. The World Health Report: World Health Organisation Pratt, T. C. 2008. Rational choice theory, crime control policy, and criminological relevance, Criminology & Public Policy , vol. 7, no. 1, 43–52 Roberts, G. 2023. BMW specialist sales to police ended after PC death , fleetnews.co.uk, https://www.fleetnews.co.uk/news/manufacturer-news/2023/01/20/bmw-specialist-sales-to-police-ended-after-pc-death (date last accessed February 24, 2024) Rossi, P. H. 1987. The iron law of evaluation and other metallic rules, Research in Social Problems and Public Policy , vol. 4, no. 1, 3–20 Sherman, L. W. 2013. Targeting, Testing and Tracking Police Services: The Rise of Evidence-Based Policing, 1975-2025, in Crime and Justice in America, 1975-2025 , Chicago, University of Chicago Press Sherman, L. W., Gartin, P. R., and Buerger, M. E. 1989. Hot spots of predatory crime: routine activities and the criminology of place, Criminology , vol. 27, no. 1, 27–55 Sherman, L., Neyroud, P. W., and Neyroud, E. 2016. The Cambridge Crime Harm Index: Measuring Total Harm from Crime Based on Sentencing Guidelines, Policing: A Journal of Policy and Practice , vol. 10, no. 3, 171–83 Simmonds, D., Ariel, B., and Harinam, V. 2023. Overcoming unreported violence using place‐based ambulance data: The case for mapping hotspots based on health data for crime prevention initiatives, Transactions in GIS , vol. 27, no. 7, 1928–41 Statistics, O. for N. 2023. Crime in England and Wales: year ending June 2023 , https://ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/bulletins/crimeinenglandandwales/yearendingjune2023 Sutherland, A., Strang, L., Stepanek, M., Giacomantonio, C., Boyle, A., and Strang, H. 2021. Tracking Violent Crime with Ambulance Data: How Much Crime Goes Uncounted?, Cambridge Journal of Evidence-Based Policing , vol. 5, nos. 1–2, 20–39 Thaler, R. H. and Sunstein, C. R. 2009. Nudge: improving decisions about health, wealth and happiness , London, UK, Penguin Weinborn, C., Ariel, B., Sherman, L. W., and Dwyer, E. O. 2017. Hotspots vs. harmspots: Shifting the focus from counts to harm in the criminology of place, Applied Geography , vol. 86, 226–44 Weisburd, D. 2015. The law of crime concentration and the criminology of place, Criminology , vol. 53, no. 2, 133–57 Weisburd, D., Groff, E. R., and Yang, S.-M. 2012. The criminology of place: Street segments and our understanding of the crime problem, in Oxford, UK, Oxford University Press Williams, S. A. 2015. “Do Visits or Time Spent in Hot Spots Patrol Matter Most? A Randomised Control Trial in the West Midlands Police” Williams, S. and Coupe, T. 2017. Frequency Vs. Length of Hot Spots Patrols: a Randomised Controlled Trial, Cambridge Journal of Evidence-Based Policing , vol. 1, no. 1, 5–21 Zimring, F. E. and Hawkins, G. J. 1973. Deterrence: The Legal Threat in Crime Control , Chicago, IL, University of Chicago Press Tables Tables 1 to 6 are available in the Supplementary Files section Additional Declarations No competing interests reported. Supplementary Files Tables.docx Cite Share Download PDF Status: Published Journal Publication published 11 Feb, 2025 Read the published version in Cambridge Journal of Evidence-Based Policing → Version 1 posted Editorial decision: Revision requested 17 Nov, 2024 Reviews received at journal 05 Nov, 2024 Reviewers agreed at journal 19 Oct, 2024 Reviews received at journal 23 Sep, 2024 Reviewers agreed at journal 18 Sep, 2024 Reviewers agreed at journal 16 Sep, 2024 Reviewers invited by journal 15 Sep, 2024 Editor assigned by journal 08 Sep, 2024 Submission checks completed at journal 14 Aug, 2024 First submitted to journal 27 Jun, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4650164","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":351134581,"identity":"a8403345-e368-4f0a-b4f3-12f5e1a1ce71","order_by":0,"name":"Tori P. A. Olphin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAu0lEQVRIiWNgGAWjYFAC5gZmBgMGORDzwANiNPAwMIK1GIO1JBCvhYEhsQHEI0qLPXtj2+OCgjvp88MOPwTaYien20DIFp6D7cYzDJ7lbrydZgDUkmxsdoCQFonENmkeg8O5G2cngLQcSNxGrJZ0w9npH0jTkiAvnUOsLWeAfgFqMdwgnVNwIMGACL+wtzcfe8zz57C8/Oz0zR8+VNjJEdQCBGxg0gCs0oCwcoQW+QbiVI+CUTAKRsEIBABLFUN6GjTaOQAAAABJRU5ErkJggg==","orcid":"","institution":"Thames Valley Police","correspondingAuthor":true,"prefix":"","firstName":"Tori","middleName":"P. A.","lastName":"Olphin","suffix":""},{"id":351134582,"identity":"11496281-d094-4d70-bd6c-525a4c9ca4fe","order_by":1,"name":"Owen Miller","email":"","orcid":"","institution":"Thames Valley Police","correspondingAuthor":false,"prefix":"","firstName":"Owen","middleName":"","lastName":"Miller","suffix":""},{"id":351134583,"identity":"1cadd9d9-da11-48b4-9393-9cbc018deef8","order_by":2,"name":"Nick Portnell","email":"","orcid":"","institution":"Thames Valley Police","correspondingAuthor":false,"prefix":"","firstName":"Nick","middleName":"","lastName":"Portnell","suffix":""},{"id":351134584,"identity":"4f234f3e-9a29-43ed-8b6e-f6be56284f29","order_by":3,"name":"Lewis Prescott-Mayling","email":"","orcid":"","institution":"Thames Valley Police","correspondingAuthor":false,"prefix":"","firstName":"Lewis","middleName":"","lastName":"Prescott-Mayling","suffix":""},{"id":351134585,"identity":"02cabe0e-6293-4da9-ac82-2199b5511b38","order_by":4,"name":"Chris Young","email":"","orcid":"","institution":"Thames Valley Police","correspondingAuthor":false,"prefix":"","firstName":"Chris","middleName":"","lastName":"Young","suffix":""},{"id":351134586,"identity":"d40148ed-8acc-470e-9dac-e3eda1ab6eca","order_by":5,"name":"Jade Hewitt","email":"","orcid":"","institution":"Thames Valley Police","correspondingAuthor":false,"prefix":"","firstName":"Jade","middleName":"","lastName":"Hewitt","suffix":""}],"badges":[],"createdAt":"2024-06-27 16:25:55","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4650164/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4650164/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s41887-024-00096-7","type":"published","date":"2025-02-11T15:57:40+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":64383424,"identity":"d2617393-f4e3-434b-9d13-154c197d4787","added_by":"auto","created_at":"2024-09-12 12:10:13","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":522896,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 1 and 2. Screenshots from the hotspots app\u003c/p\u003e","description":"","filename":"Fig12.png","url":"https://assets-eu.researchsquare.com/files/rs-4650164/v1/264ff61e90b621f9497396a3.png"},{"id":64383426,"identity":"6f81c064-005e-489d-969f-96870b3f6cf8","added_by":"auto","created_at":"2024-09-12 12:10:13","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":25945,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 3. Distribution of patrol length\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-4650164/v1/75565ae7ca915decd0ccaf97.png"},{"id":76487572,"identity":"58414a5f-d0da-4044-b071-c90ef1a1835f","added_by":"auto","created_at":"2025-02-17 16:09:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1069166,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4650164/v1/cf3bdd4e-cafa-4b4e-b082-60da86eb107a.pdf"},{"id":64383425,"identity":"fe76961c-ba91-40ed-9858-a68d2fdf4986","added_by":"auto","created_at":"2024-09-12 12:10:13","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":61950,"visible":true,"origin":"","legend":"","description":"","filename":"Tables.docx","url":"https://assets-eu.researchsquare.com/files/rs-4650164/v1/35494453a0716fc1c35caf6e.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Testing Application-Based Tasking and Hotspots Policing: A Two-in-One Randomised Trial","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThis study is concerned with randomised testing of hotspots policing, using targeted patrols of Joint Operations Unit officers to hotspots of violent crime through use of a newly developed mobile phone application in Thames Valley Police, the largest non-metropolitan policing agency in the United Kingdom.\u003c/p\u003e \u003cp\u003eViolent crime has been recognised for a long while by the World Health Organisation (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e1997\u003c/span\u003e) as a serious international health concern, with clear connections to disability, personal injury and illness. The Office for National Statistics (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) shows that there were an estimated 0.9\u0026nbsp;million violent crimes in the year ending June 2023, and that this was no significant change from the previous year, indicating a massive amount of harm that is being done to members of the UK public each year due to interpersonal violence.\u003c/p\u003e \u003cp\u003eThere are a number of mechanisms or theories that have led to the development of hotspots policing as an approach to reduce crime, and that we can take into account alongside the existing evidence to build a strong methodology with which to test hotspots policing in the Thames Valley. These mechanisms are that people can be deterred from committing crimes, and that crime is concentrated in some places more than others.\u003c/p\u003e \u003cp\u003eDeterrence theory suggests that potential offenders can be prevented from committing crimes when the perceived costs of committing the offence outweigh the perceived benefits from doing so (Zimring and Hawkins, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e1973\u003c/span\u003e). This relates to rational choice theory (Pratt, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2008\u003c/span\u003e), which suggests that people think through their offending, and make decisions about whether to commit criminal offences based on the likelihood of being caught or punished. Routine activities activity (Cohen and Felson, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e1979\u003c/span\u003e) suggests that for crime to take place, there needs to be a victim and an offender in the same place at the same time, and this needs to occur in the absence of a capable guardian. Nagin and colleagues (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) argue that increasing visibility of police officers in hotspots will create deterrent effects by increasing the perceived likelihood of apprehension. Based on all of the above, for hotspots policing to be effective at reducing violent crime, there would be a requirement for the offender to either notice a patrolling officer or be uncertain as to the lack of presence of an officer, and feel that the risk of committing such an offence would outweigh the benefit or lack of thought taken to commit it.\u003c/p\u003e \u003cp\u003eIt is likely that hotspots policing may be more difficult, and potentially less effective, for prevention of violent crime due to the prevalence of crimes involving alcohol or drugs (Duke et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), and findings that violence is more likely to occur when alcohol has been consumed than when it has not been (Hagg\u0026aring;rd-Grann et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2006\u003c/span\u003e). Alcohol or drugs may lead to decreased inhibition (Duica et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), and in turn may reduce the deterrence effect or at least a lingering deterrence effect from having seen an officer earlier in the evening. A second factor that may make non-pre-meditated violence more difficult to deter is where violence occurs due to a personal insult and immediate violent response (Cohen et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e1996\u003c/span\u003e). This fast thinking or not thinking reaction may well be less easy to deter due to a lack of consideration of consequences (Kahneman, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Additionally, many violent offences occur in buildings such as pubs or clubs, and it may be more difficult to provide a deterrent effect within buildings. This is consistent with the findings of Braga and colleagues (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) which showed that mean effect size of hotspots policing on violent crime (0.102) is lower than the mean effect size on drug (0.244), disorder (0.161) or property-related (0.124) crime outcomes. Consistent with Rossi\u0026rsquo;s (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e1987\u003c/span\u003e) Stainless Steel Law of Evaluation which suggests that the better designed the impact assessment, the more likely it is to find net impact to be zero, they also found that randomised controlled trials were associated with lower mean effect sizes (0.109) than quasi-experimental designs (0.171).\u003c/p\u003e \u003cp\u003eThe consistent finding that both crime (Sherman et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e1989\u003c/span\u003e) and crime harm (Weinborn et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) are concentrated in some areas more than others has led Weisburd (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) to suggest that there is a law of crime concentration at places, and that this occurs to a similar degree across different cities and throughout time. This remains true even in areas that have extremely high levels of crime, with the majority of the space being relatively crime free and a small percentage of the area accounting for the vast majority of the crime (Weisburd et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). This, added to the theory that capable guardians being present at a location can deter potential offenders from committing crime provides a very compelling argument for hotspots policing, as succeeding at deterring crime at locations which account for a high proportion of all crime would have a large impact on the overall level of crime in a city whilst deploying officers efficiently. There is a significant quantity of evidence (Braga et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) from hotspots policing trials in the United States which suggests that especially for drug crime, disorder and property crime, hotspots policing is associated with decreased levels of crime.\u003c/p\u003e \u003cp\u003eDespite having an extensive evidence base from the United States, we still have significant gaps in our knowledge of what works in the United Kingdom. Of the 65 eligible hotspots policing evaluations included in meta-analysis by Braga and Colleagues (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), 51 were conducted in the US, and only 4 in the UK (Ariel et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Ariel and Partridge, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Fielding and Jones, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Williams, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Since the cutoff date for that meta-analysis, these authors were able to find two more randomised trials (Basford et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Bland et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), one more publication of one of the trials in the meta-analysis (Williams and Coupe, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), and two more quasi-experimental trials (de Brito and Ariel, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Gibson et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) of hotspots policing that that have been conducted in the UK and published, so the evidence base is growing, but there are still significant evidence gaps in relation to the efficacy of hotspots. There were some interesting findings from these UK based studies that have also gone into our thinking for design of this research. Basford and colleagues found that in small (150m\u003csup\u003e2\u003c/sup\u003e) hotspots of assault, robbery and drug dealing, it was possible to reduce crime harm on days with patrols by operational support group officers, generally experienced officers who are well practiced in proactive policing. Williams and Coupe (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) found that even when total patrol time was controlled for, fewer longer (mean 2.5 patrols of 9.6 minutes) patrols were associated with lower levels of crime than hotspots with more shorter patrols (mean 5 visits averaging 5.2 minutes each). This adds further weight to Koper\u0026rsquo;s (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e1995\u003c/span\u003e) correlational findings that patrols of 10\u0026ndash;15 minutes appeared optimal for hotspots patrols to reduce crime. Ariel and colleagues (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) found that additional patrols by police community support officers, a visible patrol officer with limited powers of arrest, also led to decreased crime, suggesting that the visibility of a capable guardian may be more of a factor than the powers at the disposal of that capable guardian.\u003c/p\u003e \u003cp\u003eAriel and Partridge(2017) showed potential for a backfire effect in relation to hotspots policing and hypothesised that this may be due to identifiable patrol patterns. However, based on the level of power found in other hotspots trials, the lack of a crossover design in this trial may have meant that it was simply underpowered and whilst we will remain aware of the potential for a backfire effect, these findings are not of major concern in design of this trial. Unfortunately, whilst some of the other studies tested interesting concepts such as reducing patrol(Gibson et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) and the effect of feedback and tracking on delivery of patrols (de Brito and Ariel, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), both of these trials were quite underpowered, so it would not be possible to draw generalisable conclusions from them.\u003c/p\u003e \u003cp\u003eWhilst there have been strong effects reported by Bland and colleagues (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), who found that it was possible to prevent serious violence across a large area with minimal amounts of foot patrol, this study was conducted during periods of lockdown in the UK. The presence of lockdown restrictions is likely to have altered both the underlying level of crime due to reductions in the numbers of people present in public, and the general presence of police officers on routine patrol. In addition, when people were out in public, they had to be outdoors to congregate which will make it much easier for officers to be seen. Therefore it is possible that these findings were an anomaly due to it being easier to see police officers, easier for police officers to see incidents occurring, and a lower baseline level of patrol at the time.\u003c/p\u003e \u003cp\u003eIt is therefore suggested that the best existing evidence would be for around three patrols per day each lasting approximately 15 minutes, in small hotspots where the officer can be seen by anyone in the hotspot for at least part of the patrol if not the majority of it and that this would be a good place to start testing, given that we do not currently know what works best for hotspots policing in non-metropolitan areas of the United Kingdom, especially in policing areas with a massive land area.\u003c/p\u003e \u003cp\u003eDespite the high level of evidence surrounding hotspots policing, in the UK there are still many unanswered questions and, in areas that have implemented hotspots policing, there are often large additional costs such as GPS trackers which may prohibit large scale implementation. Consistent with other researchers (Basford et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Bland et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), at the start of this trial we were not aware of any UK police agency that had a force-wide patrol strategy to target hotspots of violence that they were tracking delivery of on a daily basis. We were also not aware of any test involving implementation of place-based policing using a mobile phone application, that linked securely into police systems, as the tasking mechanism; something that allows for much more cost-effective implementation, and which would remove the technological barriers to implementation of hotspots policing raised by Ariel (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), and allow for tracking of delivery, argued to be essential in any implementation (Sherman, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), to be automated. Therefore, it is the aim of this research to design an implementation that is the best possible test of the existing best-evidence in a way that can be rolled out to business-as-usual across the entire force area for low to no additional cost through concurrent testing of a new hotspots policing mobile phone app.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThames Valley Police is the largest non-metropolitan police force in the United Kingdom, covering over 2,200 square miles and three counties; Berkshire, Buckinghamshire and Oxfordshire. This massive area and range of urban and rural areas meant that careful thought needed to go into how additional policing could best be delivered to hotspots spread across the Thames Valley, especially given the large distances between hotspots that would make it incredibly difficult to have dedicated hotspots patrol resources moving between hotspots.\u003c/p\u003e\n\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eChoice of Operational Resource\u003c/h2\u003e\n \u003cp\u003eLimiting the experiment to individual parts of the force area would not have elicited sufficient locations to provide analytic power capable of showing a result. To deliver consistent policing of hotspots across the entirety of the Thames Valley, using officers who were equipped to travel to the hotspots and whose patrols could be managed to avoid treatment of hotspots on days where those hotspots were in the control group, it was decided that the Joint Operations Unit (JOU) would be the most appropriate resource for conducting patrols.\u003c/p\u003e\n \u003cp\u003eThe JOU is a shared resource with Hampshire Constabulary, and comprises roads policing unit, firearms capability, operational support, canine unit, and mounted unit. The JOU is a flexible resource, used to conducting disparate types of policing operation, and they are well equipped to travel across the Thames Valley to conduct policing activities. In addition, it was seen as essential that we were able to manage the resources effectively and ensure that all areas were receiving comparable policing of hotspots. The JOU allowed us to use one command structure to achieve this, avoiding large amounts of liaison between different command structures as would be the case if conducting patrols across different policing areas using their own local resources. Patrolling was completed during both downtime within the JOU\u0026rsquo;s core shift time (i.e. time between attending incidents), and in periods of funded overtime offered to JOU officers to focus specifically on the violent crime hotspots.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\n \u003ch2\u003eSelection of hotspots\u003c/h2\u003e\n \u003cp\u003eThe randomised controlled trial was planned for implementation during the period from September 2021 to March 2022, and to avoid introducing seasonal hotspots that would not be relevant to the time period of the RCT, the data used to build hotspots was restricted to occurrences that took place only during these months from September 2016 to March 2020. Data from 2020 to 2021 was not used, due to the unknown effects of lockdowns during the covid-19 pandemic.\u003c/p\u003e\n \u003cp\u003eA visual search of the locations of incidents was conducted, and the locations of serious violent occurrences was consistent with that of violent offences with less serious outcomes. It was decided that all violent acts, from public order offences all the way up to murder, would be incorporated into the data used to build hotspots, so that locations would not be skewed by small numbers of incidents which would not be preventable during the test period. Domestic violence offences were removed from the dataset, as these were deemed be much less preventable through patrol, due to high prevalence of these crimes occurring inside private dwellings.\u003c/p\u003e\n \u003cp\u003eAn initial dataset of incidents (n\u0026thinsp;=\u0026thinsp;59,103) was examined for spatial quality in terms of whether the location was able to be plotted on a map using either XY coordinates or postcode. Where the location was able to be plotted using XY coordinates or postcode, these incidents were geocoded and mapped (n\u0026thinsp;=\u0026thinsp;58,959).\u003c/p\u003e\n \u003cp\u003eTo control for likely differences in violent crimes that occur in the daytime and in the evenings, the data were split into two periods; days (from 08:00 to 19:59, n\u0026thinsp;=\u0026thinsp;35,494) and nights (from 20:00 to 07:59, n\u0026thinsp;=\u0026thinsp;23,465). These were the final datasets for identification of hotspots. Once hotspots were identified, it became apparent that all the days\u0026rsquo; and nights\u0026rsquo; hotspots were different areas, and this supported the decision to split the data and identify spatio-temporal hotspots.\u003c/p\u003e\n \u003cp\u003eTo allow for us to test whether a \u0026lsquo;standard\u0026rsquo; size of hotspot (one where an officer can be seen in all of the hotspot during a patrol) is an effective mechanism for hotspots policing in the Thames Valley, a tessellation of hexagons was created for the entire Thames Valley Police area. Crime levels in these hexagons were examined to ensure that the size of hotspot could be kept as small as possible. This would ensure that the entire hotspot could be patrolled during a fifteen-minute patrol, whilst still containing sufficient crime that a reduction could realistically be detected through testing. This reached a balance where hexagons with edge length of 150m were selected as being appropriate in both measures. This ensured that sufficient crimes were included in the highest concentration hotspots, and that they could be patrolled in a manner that officers could be seen from the majority of the hotspot while they were patrolling, maximising the impact of patrols. This resulted in just under 145,000 hexagons across the force area. Incidents were aggregated into the hexagons, and the top 50 hotspots based on crime count for days and nights were retained for manual examination.\u003c/p\u003e\n \u003cp\u003eHexagons which included hospitals, prisons and schools were removed manually, as the majority of crimes in these hotspots are likely to have occurred indoors, and would not be preventable through patrols. A seventy-five metre buffer zone was applied to each hotspot, so that cross-contamination of patrols could be minimised. Sanity checks were conducted to ensure that these hotspots were enduring (consistent over time), that they remained consistent when examined using different methods of hotspot identification (clustering and optimised hot spot analysis), and that they did not miss high concentrations of crime harm. Mapping against Cambridge crime harm index scores (Sherman et al., \u003cspan class=\"CitationRef\"\u003e2016\u003c/span\u003e), and year by year examination did not change the locations of the selected hexagons, and there were no additional areas that appeared to be as appropriate as the ones selected.\u003c/p\u003e\n \u003cp\u003eThis was kept in sight whilst the remaining hotspots were reviewed and manually manipulated (rotated and/or nudged spatially) to ensure that; there were no overlapping buffer areas, areas such as dual carriageways which could not be patrolled were removed, and the incident inclusion was maximised. This spatial nudging was done to ensure that the hotspots were in the most appropriate places, not just where the tessellation had originally placed them, and was done with view of all crime incidents on the map at the same time. Aggregated totals were recalculated, and the research team reviewed all the final hexagon locations with tactical leads to ensure that the hotspots made sense from their point of view. Forty-five hotspots were identified for the experiment (days n\u0026thinsp;=\u0026thinsp;19; nights n\u0026thinsp;=\u0026thinsp;26). This was accepted by tactical decision makers as being reasonable for delivery of patrols.\u003c/p\u003e\n \u003cp\u003ePower analysis calculations were conducted to establish the likely case count that would be required to show an effect using previously found effect size of 0.102 for impact of hotspots policing on violent crime (Braga et al., \u003cspan class=\"CitationRef\"\u003e2019\u003c/span\u003e), and likely standard deviation of control hexagons of 0.01. According to these calculations, the expected number of cases required in each group in order to show an effect as calculated above, at the mean positive effect size of previous studies is 2021 for power of 0.9, and 1510 for power of 0.8. That means that between 3020 and 4042 total treatment or control periods would be required.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003eTasking, targeting and tracking via a mobile phone application\u003c/h2\u003e\n \u003cp\u003eHaving identified a roaming resource to carry out the hotspots patrols as additional policing and the hotspots that they would be tasked to patrol, it was necessary to design a mechanism by which officers could be briefed on the areas to visit, and hotspots were randomised and correctly displayed to officers. There was also a necessity to record how many patrols were conducted to track the patrols. This research team is aware that in other areas of the UK, control room resources are used to direct officers into the hotspots, and this was considered as an option. However, the view of this research team is that this should not be a test of what is possible if we throw money at a problem, but rather what solution might be possible that would deliver a long-lasting effect, and that could be made into a business-as-usual delivery, with no additional cost. This would mean that it can be scaled up to deliver a lasting improvement in service and efficiency, rather than something that would disappear if funding were no longer available.\u003c/p\u003e\n \u003cp\u003eTo satisfy the requirements of targeting the appropriate hotspots, tracking the visits when they occur, and creating a tasking mechanism that is sustainable and scalable, members of this research team designed a hotspots patrol application using Microsoft Power Apps. This application was logged into by officers on their force mobile phones, and presented them the hotspots that were allocated for patrol that day or night. It showed them an overview map to help them get to the location, and a detailed map of what was in the hotspot, as well as a postcode for satellite navigation and details of any warning markers to look out for. Figures \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e and 2 below shows two screenshots from the live version of the app, as it would be seen by officers.\u003c/p\u003e\n \u003cp\u003eWhen officers started their patrol in a hotspot, they click a button to start the recording of the patrol, and a timer popped up to show them how long they had been actively patrolling. This turned amber when it got to 13 minutes, then green at 15 minutes. This was designed to add an element of a nudge effect (Thaler and Sunstein, \u003cspan class=\"CitationRef\"\u003e2009\u003c/span\u003e) to work in addition to training about what length a patrol should be, and encourage patrols of a length that would match that suggested by Koper (\u003cspan class=\"CitationRef\"\u003e1995\u003c/span\u003e) as being the optimal length to see a deterrent effect that lasts. When the patrol finished, they clicked again, and then used sliders to provide details of patrol activities carried out (positive engagement, stop and search, and arrests).\u003c/p\u003e\n \u003cp\u003eThis is therefore different from most other hotspots policing interventions which are quite resource intensive for management of the intervention. This experiment is testing both whether tasking officers using an application on their mobile devices results in additional patrols, and whether this in turn creates a change in the amount of violent crime in small hexagonal hotspots within Thames Valley Police when roaming joint operations unit officers are tasked to patrol them.\u003c/p\u003e\n \u003cp\u003eThe data from the application was fed through a Microsoft Azure environment to a Power BI dashboard where it was updated every two hours so our research team could track delivery of treatment during the experiment. This dashboard was reviewed with our tactical leads twice a week to ensure that treatment fidelity was as high as possible. A backup mechanism using Microsoft Forms was also created in case of failure of the app, though this was fortunately not needed during the delivery of the trial.\u003c/p\u003e\n \u003cp\u003eIn addition, archive data from airwave radios was collected for the hotspot areas so that the level of overall patrol in the hotspots could be examined, along with the ratio of pings from JOU radios between the test and control hotspots.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\n \u003ch2\u003eTesting: how the experiment was conducted\u003c/h2\u003e\n \u003cp\u003eThe randomisation was a cross-randomised design, where hotspots were randomised each day to one of treatment or control, and each hotspot will therefore appear as treatment and control dependent on the day or night in question. Cases were pre-randomised using the sample() function in R (Ihaka and Gentleman, \u003cspan class=\"CitationRef\"\u003e1996\u003c/span\u003e). Pre-randomisation was used so that the randomisation could be incorporated into the app and the dashboards. However, the actual randomisation was blind to both researchers and officers until the point that the application displayed the treatment hotspots on the day. The randomisation was a simple randomisation each day to ensure that the number of treatment hotspots are consistent (9/10 out of 19 for days, and 13 out of 26 for nights), This would keep the experiment within bounds that were agreed with Joint Operations Unit in relation to the number of hotspots they believed they would be able to effectively patrol.\u003c/p\u003e\n \u003cp\u003eOfficers were given guidance at the beginning of the experiment in how to use the app, and what was expected of them. They were briefed that repeated patrols (3 per hotspot per day) were desirable, that patrols should last between 13 and 20 minutes, and that they should be concentrating their patrols on the hotspots that were allocated through the application, not based on previous days.\u003c/p\u003e\n \u003cp\u003eThe test period was pre-defined as being between 08/11/2021 and 31/03/2022 inclusive. This time period was a constraint of the experiment, due to funding windows. It was decided at the start of the experiment that the outcomes should be blind to the research team until the experiment was concluded, in order that knowledge about how crime was being affected could not be used in any way to alter the integrity of the trial.\u003c/p\u003e\n \u003cp\u003eThree factors arose at one point during the RCT which would have severely limited the ability to provide resources for the experiment between 19/12/2021 and 03/01/2021 inclusive; Omicron Covid-19 variant limited overall staffing levels, overtime was paused meaning no additional staffing, and there were some issues with vehicles (Roberts, \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e) which limited patrols by JOU units. Therefore, it was decided that the period between 19/12/2021 and 03/01/2021 inclusive would not be part of the RCT, and no measurement would be taken from this period. No analysis of crime levels was done prior to making this decision, so the researchers were blind to the ongoing performance of the trial, and the decision was made in advance of the problems occurring due to identification of the potential issues during our bi-weekly planning meetings. It was therefore believed to be fair and reasonable to make this decision at the time.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n \u003ch2\u003eAnalysis plan\u003c/h2\u003e\n \u003cp\u003eThis experiment is examining two different outcomes; whether the tasking of officers via a mobile phone application results in an increase in patrol time, and in turn whether this then has a knock-on effect on crime.\u003c/p\u003e\n \u003cp\u003eTo answer the first part, data from the application will be used to show the mean number of patrols per day and give an overview of the level of patrol that was added through development of the application. In addition to this, airwave pings will be examined when they occur within the hotspots. The airwave pings also record the speed of travel at the time, and since we are looking for patrols that have the potential to deter criminal activity it was decided that pings over 50km/h should not be counted. Therefore, the mean pings in the hotspots per day will be examined both for joint operations unit officers, and for all other officers to provide a baseline of the activity in those areas.\u003c/p\u003e\n \u003cp\u003eTo answer the question of whether this approach has then led to a change in the level of crime, three different types of crime will be examined.\u003c/p\u003e\n \u003cp\u003eFirstly, violent crime is defined for this research as any offence against the person where violence is used, or feared that it will be used. Therefore, this includes all offences from S.4 Public Order Act, up to and including Homicide.\u003c/p\u003e\n \u003cp\u003eThe second category adds in sexual offences to the above group, as the level of violence used in a common assault and a sexual assault may well be the same, yet sexual assaults are not included in most examinations of violent crime. This may therefore reduce the visibility of women and girls as victims of public place offending. To avoid this, all offences of sexual assault, with or without penetration, and rape will be included in a violent and sexual offences category.\u003c/p\u003e\n \u003cp\u003eThirdly, total recorded crime, including all offences, will be examined to establish whether there is an overall impact over and above that seen in terms of violent crime.\u003c/p\u003e\n \u003cp\u003eAll three of the above crime types will be examined in terms of their prevalence, or whether there is one or more crime of that category in that location on that day, the mean count per day of that category, and the mean harm as measured by the Cambridge crime harm index (Sherman et al., \u003cspan class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003eAll of the above analyses of both change in patrol and change in crime will be conducted as an overall combination of all days and nights hotspots, and will also then be broken down to examine days and nights independently to establish whether there are any differential levels of impact at different times of day. The analyses are based on a pure intent to treat approach, and therefore the analysis will be conducted based on the randomised allocation, regardless of level of patrol that was conducted. The intent to treat approach is used because this is a test of an implementation of an intervention in a live environment, and it is therefore not only important to establish whether it works when implemented appropriately, but also how it works in the real world (Detry and Lewis, \u003cspan class=\"CitationRef\"\u003e2014\u003c/span\u003e), as the implementation of a low-contact tasking mechanism for officers via a mobile phone application is what is being tested, and it would not be a fair test if analysis were restricted only to times where it was used correctly.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003eFindings\u003c/h2\u003e\n \u003cp\u003eIn this research, two main questions were asked; firstly, does tasking officers to attend hotspots of violent crime, using a mobile phone application, result in more patrols? Secondly, when hotspots of violence were identified using best practice found in previous research, and officers are patrolling them, does this result in a reduction in violent crime?\u003c/p\u003e\n \u003cp\u003eAs can be seen from Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e, there was an average of 1.255 patrols of at least 13 minutes per treatment hotspot per day by officers using the application, totalling over 3,500 additional patrols. To assess whether there was a change in actual time in the hotspots, the recorded location pings from airwave radios were compared, and there were significantly more airwave pings by joint operations unit officers in the treatment hotspots than in the control hotspots, t(5162.8)\u0026thinsp;=\u0026thinsp;18.76, p\u0026thinsp;\u0026lt;\u0026thinsp;0.05. This was an increase of 93% over the control hotspots, increasing from 5.89 pings per control hotspot per day to 11.39 pings per treatment hotspot per day. Joint Operations Unit officers only account for a small proportion (around 10%) of frontline officers in Thames Valley Police, so whilst this did also increase the overall amount of policing activity when all officers were included, this increase was non-significant.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eWhilst the above differences in patrols and airwave pings were seen in both days (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e) and nights (Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e) hotspots, the increase in patrol was greater for days hotspots than nights hotspots. There were 2.01 patrols on average per treatment hotspot per day for days hotspots, but only 0.713 for nights hotspots. The mean count of airwave pings increased from 4.06 to 13.13 pings per hotspot per day for days hotspots, t(1835.9)\u0026thinsp;=\u0026thinsp;21.864, p\u0026thinsp;\u0026lt;\u0026thinsp;0.05, and increased from 7.24 to 10.11 pings per hotspot per day for nights hotspots, t(3124.4)\u0026thinsp;=\u0026thinsp;7.168, p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003eAdditionally, it was found that the vast majority of patrols exceeded the desired 15 minutes as demarcated by the change in colour of the timer in the app. Figure \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eshows the distribution of patrols of different lengths, which can clearly be seen to be a non-normal distribution.\u003ctable id=\"Tab2\" border=\"1\"\u003e\u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eIn relation to the levels of crime seen in the treatment and control hotspots, Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e shows that whilst there appears to have been a decrease in the prevalence of crime in the hotspots, this was not a significant finding. For the sake of transparency, all analyses are shown below. From this, it can be seen that there was on average only one crime of any type every two days, and this decreased for violent crime to one every 5 days in the areas that were being patrolled.\u003c/p\u003e\n \u003cp\u003eTable 5 and Table 6 show that these findings were consistent across days and nights, with violent crime being more prevalent during the nights hotspots overall, but it appears that despite the hotspots being created in a manner consistent with prior research, and these being the areas of highest concentration of violent crime, the overall prevalence of crime in Thames Valley hotspots created in this way was still low.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003eThese findings offer many options for development of this approach, and indicate that it is possible to develop an extremely cost-effective tasking mechanism for targeting of geographical locations.\u003cp id=\"Tab4\" border=\"1\"\u003e\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis randomised trial of the implementation of hotspots policing using a mobile phone app to direct police activity has demonstrated that it is possible to remove many of the barriers to implementation of hotspots policing using a well-designed application that is created with use by officers in mind.\u003c/p\u003e \u003cp\u003eThe application allowed for officers to be tasked and tracked, on a business-as-usual basis, in a way that did not require additional resourcing that wasn\u0026rsquo;t delivering the patrol. An alternative would be to use supervisory or control room staffing to direct the patrols in real time. This can become a cost-burden for police services, without adding patrol value. It is therefore believed that this method shows a good compromise between in-the-moment resource allocation and tracking. However, it is acknowledged that there can be other benefits of using additional tasking resources; they would allow for tracking feedback to be given in the moment of the patrol if needed, which could improve quality of patrols. The regular tracking of patrols was conducted using airwave radio data, and this provided a balance between low cost and no resource requirement, and up to date feedback for officers. It is also possible that the application could be developed further to allow for tasking of other types of patrol, for problem-oriented policing activity and tracking of actions taken, as well as for other non-immediate taskings of which police officers have many, so it is a scalable solution, with little-to-no additional cost requirement. There was an error caused by application data not refreshing if the app was not closed overnight which led to a small number of patrols of hotspots on control days. This is likely to have been a limitation in the mechanism used for access of the application, as they were not available to be installed on phones, and had to be accessed via a web application. This should be fixed for future versions to ensure that there is no bleed of treatment into control areas.\u003c/p\u003e \u003cp\u003eThe application has been released for use by other UK police forces, and is also being developed further to ensure that this tasking mechanism can be used for operational deployments across Thames Valley Police. Both of these will ensure that the aim of the project which was to develop and deliver sustainable mechanisms for tasking of officers to hotspots of crime is successful. This is a technology that is easily scalable, and can be applied in many different aspects of policing.\u003c/p\u003e \u003cp\u003eThe nudge effect (Thaler and Sunstein, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2009\u003c/span\u003e) of having the timer change colour at 15 minutes appears to have been extremely beneficial, but needs specific additional research. It also lends credence to using this type of approach in other areas of delivery in the public sector. However, there is a need to be incredibly careful to avoid backfire effects often associated with gaming of things that are seen as targets (Grace, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWhilst overall, the use of an app appears to be an effective mechanism for tasking of police resources, there was a larger increase in airwave pings, which can loosely be interpreted as police presence, on days than on nights, indicating that there may be a limitation on resource availability when the app is used later in the day. This is likely to have been compounded slightly due to the timings of overtime shifts which were offered between 2pm and 10pm each day. However, this was a substantial increase in directed policing by resources that would not have been in these areas otherwise (both through business-as-usual resourcing patrolling during downtime, and in directing overtime resources to specific tasks or locations). It is therefore likely that this approach can be expanded out to all business-as-usual resources to allow improved tasking at a whole-force level. It may be possible to improve the tasking further, if the application can be developed to show current location in relation to the hotspot. This may act as a further nudge to ensure that officers keep patrolling the hotspot while they are allocated to. There was an overestimation of how much patrol could reliably be provided, especially on night shifts, and this did impact upon what was possible. It would be beneficial for all forces to establish a baseline of activity for officers who are to be used for hotspots policing, prior to launch of hotspots policing strategies.\u003c/p\u003e \u003cp\u003eIt also appears that, whilst non-significant, this approach of using joint operations unit officers to patrol small hotspots has led to a reduction in prevalence of total crime and violent crime. This benefit was minimal however, likely due to the low rate of crime commission in those hotspots, as low base rates of crime present significant challenges to evaluation of hotspots policing (Hinkle et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). It was useful to commence testing of hotspots policing using small micro-hotspots, as a test the historic hotspots literature for an area such as Thames Valley Police. However, whilst these hotspots did allow for the entire area to be patrolled by a single officer within 15 minutes, there is a compelling argument for testing of larger hotspots which would need to be contiguous areas that were straightforward to patrol. The expansion of hotspot size would then allow for the quantity and rate of crime in each hotspot to be sufficient that it would be reasonable to expect one crime per day in a hotspot, meaning that there are no days where patrol was conducted but there would not have been a crime in any case. In this study, the rate of crime was much lower than this, and this likely contributes to the lack of significant decreases in crime. The maximum size of these hotspots still needs to allow for officers to be seen across the entire area, as if officers are not seen, they will not create a deterrence effect. It may also be valuable to assess the impact of patrols which enter shops and licensed premises, to increase visibility. This, along with assessment of the quality of patrol, level of visibility of officers, and proactivity displayed might also play a part in the effectiveness or lack thereof in studies of this type. This was not possible for us to assess, though this would be beneficial where possible in future work. When creating hotspots, the level of crime present in the hotspots needs to be preventable by an occasional resource, at the same time as there needs to be a way in which the patrol or action taken could realistically lead to impact on the crime that is desired to be prevented; if the patrol is not visible, and actively patrolling, it is unlikely to impact on the commission of crime.\u003c/p\u003e \u003cp\u003eThere are limitations in relation to hotspot creation, as it is not currently possible to identify which crimes have occurred indoors and outdoors. This makes it more difficult to base the hotspots solely on crimes that are likely to be preventable, and it would be of benefit for policing agencies to be able to differentiate between crimes committed indoors and outdoors, and in public and private places. An additional way to improve the development of hotspots would be to incorporate ambulance records for incidents without a corresponding crime incident into the data used to build the hotspots. Sutherland and colleagues (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) found that approximately 90% of ambulance records for violent crime-related injury attendances did not have a corresponding police record, and that this was even lower for emergency department records. Their findings suggested that adding medical data to police data could increase levels of recorded violence by as much as 15 to 20%, and this in turn could aid the building of hotspots. A further analysis was conducted in the Thames Valley (Simmonds et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), which found that approximately 80% of ambulance records for violence did not have a corresponding police record within 100m of the ambulance attendance site.\u003c/p\u003e \u003cp\u003eWhilst the findings of this research are incredibly promising in relation to crime prevention, and have demonstrated that use of an application is a sound method for tasking patrols to hotspots, the hotspot size and crime density was not optimal. The use of experienced proactive officers is suspected to have assisted in the success of this trial, so it is necessary to test how this method would work at a whole-force level, with all officers being able to access the app and patrol larger hotspots ensuring higher base rates of crime. These changes are being incorporated into a second randomised trial which will be rolled out across the Thames Valley, allowing for an approach to be built where results can be learned from and acted upon.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis research has demonstrated that it is possible to direct police patrols using a mobile phone-based application, and that this produces an incredibly cost-effective mechanism for tasking of police resources, particularly when directing them to specific geographic areas. There are many benefits, that have already been realised in Thames Valley Police, from this; firstly, it is possible to direct officers reliably to areas where additional policing is needed, without additional cost for control room staff, and without additional impact on supervisory staff. It is also possible to use airwave data, examined in an automated manner in the cloud, to track the delivery of tasked activity and as demonstrated here, to track experimental delivery. By using a mechanism such as this app, it is also possible to increase patrol by existing resources, increasing delivery of policing activity using business as usual resources without increasing cost. This was a key driver in the development of this approach, as it is these researchers\u0026rsquo; opinion that we need to drive towards embedding best practice within existing mechanisms of policing, ideally designing research in a way that it can be delivered without additional cost and is realistically designed to allow implementation exactly as tested. It is the lead researcher\u0026rsquo;s strong belief that it is better to deliver the best test of something that can be implemented, than a perfect scientific trial of something that only works in laboratory conditions.\u003c/p\u003e \u003cp\u003eHowever, it has also shown that the optimal mechanism for hotspots policing in violent crime hotspots differs somewhat between different types of policing area, showing a dramatically smaller effect size for treatment of hotspots in a non-metropolitan police force than might have been expected. It is hypothesised that a different style of hotspots, with larger contiguous areas containing a slightly lower concentration but larger overall frequency and count of violent crime would be more effective in this type of police area. This approach is being developed for further testing in a follow-up experiment designed to examine the optimal mechanisms for delivery of hotspots policing in non-metropolitan areas with lower frequency of crime than in large cities. The application has been developed further and as well as being expanded out for use by all officers in this force, it has also been made available to other police forces in the UK so that they may also benefit from a revolutionary way of tasking policing resources to geographical areas. This will allow for a dramatic reduction in cost, and increase in reliability for delivery of hotspots policing approaches.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding and Support\u003c/strong\u003e: This would not have been possible without funding and support from Home Office GRIP Programme, and the strategic and operational support from Thames Valley Police. Any findings, opinions, conclusions, or recommendations expressed herein represent those of the authors and do not necessarily reflect the views of Thames Valley Police, or UK Home Office.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eT.O was the lead researcher and oversaw design, tracking and analysis of the experiment, and wrote the main manuscript text, figures and tables. O.M and L.P-M designed the experiment with T.O, and ensured delivery was maintained across Thames Valley Police. O.M wrote the application that was used for tasking, and all authors tested the application, both for technical delivery and geographical accuracy. N.P. is the geographer who conducted analysis of crime data and determined the location and spatio-temporal definition of the hotspots. C.Y and J.H led delivery of the patrol-related project and were responsible for liaison with command teams and local policing areas. All authors reviewed the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis would not have been possible without funding and support from Home Office GRIP Programme, and the strategic and operational support from Thames Valley Police. We are grateful to the officers from the Joint Operations Unit at Thames Valley Police and Hampshire \u0026amp; Isle of Wight Constabulary who conducted these patrols. Any findings, opinions, conclusions, or recommendations expressed herein represent those of the authors and do not necessarily reflect the views of Thames Valley Police, or UK Home Office.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are available from Thames Valley Police, but restrictions apply to the availability of these data, which are derived from sensitive operational data and so are not publicly available. Summary outputs of the data used, and the code used to analyse these, is available from the authors upon reasonable request and with the permission of Thames Valley Police. Norwegian Social Research (NOVA), but restrictions apply to the availability of these data, which were used under licence for the current study and so are not publicly available. The data are, however, available from the authors upon reasonable request and with the permission of Norwegian Social Research (NOVA).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAriel, B. 2023. Implementation issues with hot spot policing, \u003cem\u003eInternational Journal of Law, Crime and Justice\u003c/em\u003e, vol. 75, 100629\u003c/li\u003e\n\u003cli\u003eAriel, B. and Partridge, H. 2017. Predictable Policing: Measuring the Crime Control Benefits of Hotspots Policing at Bus Stops, \u003cem\u003eJournal of Quantitative Criminology\u003c/em\u003e, vol. 33, no. 4, 809\u0026ndash;33\u003c/li\u003e\n\u003cli\u003eAriel, B., Weinborn, C., and Sherman, L. W. 2016. \u0026ldquo;Soft\u0026rdquo; policing at hot spots\u0026mdash;do police community support officers work? A randomized controlled trial, \u003cem\u003eJournal of Experimental Criminology\u003c/em\u003e, vol. 12, no. 3, 277\u0026ndash;317\u003c/li\u003e\n\u003cli\u003eBasford, 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, \u003cem\u003eCambridge Journal of Evidence-Based Policing\u003c/em\u003e, vol. 5, nos. 3\u0026ndash;4, 119\u0026ndash;33\u003c/li\u003e\n\u003cli\u003eBland, 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, \u003cem\u003eCambridge Journal of Evidence-Based Policing\u003c/em\u003e, vol. 5, nos. 3\u0026ndash;4, 93\u0026ndash;118\u003c/li\u003e\n\u003cli\u003eBraga, A. A., Turchan, B., Papachristos, A. V., and Hureau, D. M. 2019. Hot spots policing of small geographic areas effects on crime, \u003cem\u003eCampbell Systematic Reviews\u003c/em\u003e, vol. 15, no. 3\u003c/li\u003e\n\u003cli\u003eBrito, C. de and Ariel, B. 2017. Does Tracking and Feedback Boost Patrol Time in Hot Spots? Two Tests, \u003cem\u003eCambridge Journal of Evidence-Based Policing\u003c/em\u003e, vol. 1, no. 4, 244\u0026ndash;62\u003c/li\u003e\n\u003cli\u003eCohen, L. E. and Felson, M. 1979. Social Change and Crime Rate Trends: A Routine Activity Approach, \u003cem\u003eAmerican Sociological Review\u003c/em\u003e, vol. 44, no. 4, 588\u0026ndash;608\u003c/li\u003e\n\u003cli\u003eCohen, D., Nisbett, R. E., Bowdle, B. F., and Schwarz, N. 1996. Insult, Aggression, and the Southern Culture of Honor: An \u0026ldquo;Experimental Ethnography,\u0026rdquo; \u003cem\u003eJournal of Personality and Social Psychology\u003c/em\u003e, vol. 70, no. 5, 945\u0026ndash;60\u003c/li\u003e\n\u003cli\u003eDetry, M. A. and Lewis, R. J. 2014. The Intention-to-Treat Principle: How to Assess the True Effect of Choosing a Medical Treatment, \u003cem\u003eJAMA\u003c/em\u003e, vol. 312, no. 1, 85\u0026ndash;86\u003c/li\u003e\n\u003cli\u003eDuica, L., Dragulescu, V., and Pirlog, M. 2020. NEUROBIOLOGICAL ELEMENTS OF HOPELESSNESS AND IMPULSIVITY IN SUICIDAL BEHAVIOR, \u003cem\u003eInternational Journal of Medical Reviews and Case Reports\u003c/em\u003e, vol. 4, no. Reports in Clinical Medicine and, 1\u003c/li\u003e\n\u003cli\u003eDuke, A. A., Smith, K. M. Z., Oberleitner, L. M. S., Westphal, A., and McKee, S. A. 2018. Alcohol, Drugs, and Violence: A Meta-Meta-Analysis, \u003cem\u003ePsychology of Violence\u003c/em\u003e, vol. 8, no. 2, 238\u0026ndash;49\u003c/li\u003e\n\u003cli\u003eFielding, M. and Jones, V. 2012. \u0026lsquo;Disrupting the optimal forager\u0026rsquo;: predictive risk mapping and domestic burglary reduction in Trafford, Greater Manchester, \u003cem\u003eInternational Journal of Police Science \u0026amp; Management\u003c/em\u003e, vol. 14, no. 1, 30\u0026ndash;41\u003c/li\u003e\n\u003cli\u003eGibson, C., Slothower, M., and Sherman, L. W. 2017. Sweet Spots for Hot Spots? A Cost-Effectiveness Comparison of Two Patrol Strategies, \u003cem\u003eCambridge Journal of Evidence-Based Policing\u003c/em\u003e, vol. 1, no. 4, 225\u0026ndash;43\u003c/li\u003e\n\u003cli\u003eGrace, S. 2022. The perverse impact of performance measures on policing: lessons from the rise and fall of out of court disposals, \u003cem\u003ePolicing and Society\u003c/em\u003e, vol. 32, no. 2, 200\u0026ndash;220\u003c/li\u003e\n\u003cli\u003eHagg\u0026aring;rd‐Grann, U., Hallqvist, J., L\u0026aring;ngstr\u0026ouml;m, N., and M\u0026ouml;ller, J. 2006. The role of alcohol and drugs in triggering criminal violence: a case‐crossover study*, \u003cem\u003eAddiction\u003c/em\u003e, vol. 101, no. 1, 100\u0026ndash;108\u003c/li\u003e\n\u003cli\u003eHinkle, J. C., Weisburd, D., Famega, C., and Ready, J. 2013. The Problem Is Not Just Sample Size, \u003cem\u003eEvaluation Review\u003c/em\u003e, vol. 37, nos. 3\u0026ndash;4, 213\u0026ndash;38\u003c/li\u003e\n\u003cli\u003eIhaka, R. and Gentleman, R. 1996. R: A Language for Data Analysis and Graphics, \u003cem\u003eJournal of Computational and Graphical Statistics\u003c/em\u003e, vol. 5, 299\u0026ndash;314\u003c/li\u003e\n\u003cli\u003eKahneman, D. 2011. \u003cem\u003eThinking, Fast and Slow\u003c/em\u003e, New York, Farrar, Straus and Giroux\u003c/li\u003e\n\u003cli\u003eKoper, C. S. 1995. Just enough police presence: reducing crime and disorderly behavior by optimizing patrol time in crime hot spots, \u003cem\u003eJustice Quarterly\u003c/em\u003e, vol. 12, no. 4, 649\u0026ndash;72\u003c/li\u003e\n\u003cli\u003eNagin, D. S., Solow, R. M., and Lum, C. 2015. Deterrence, Criminal Opportunities, and Police, \u003cem\u003eCriminology\u003c/em\u003e, vol. 53, no. 1, 74\u0026ndash;100\u003c/li\u003e\n\u003cli\u003eOrganisation, W. H. 1997. The World Health Report: World Health Organisation\u003c/li\u003e\n\u003cli\u003ePratt, T. C. 2008. Rational choice theory, crime control policy, and criminological relevance, \u003cem\u003eCriminology \u0026amp; Public Policy\u003c/em\u003e, vol. 7, no. 1, 43\u0026ndash;52\u003c/li\u003e\n\u003cli\u003eRoberts, G. 2023. \u003cem\u003eBMW specialist sales to police ended after PC death\u003c/em\u003e, fleetnews.co.uk, https://www.fleetnews.co.uk/news/manufacturer-news/2023/01/20/bmw-specialist-sales-to-police-ended-after-pc-death (date last accessed February 24, 2024)\u003c/li\u003e\n\u003cli\u003eRossi, P. H. 1987. The iron law of evaluation and other metallic rules, \u003cem\u003eResearch in Social Problems and Public Policy\u003c/em\u003e, vol. 4, no. 1, 3\u0026ndash;20\u003c/li\u003e\n\u003cli\u003eSherman, L. W. 2013. Targeting, Testing and Tracking Police Services: The Rise of Evidence-Based Policing, 1975-2025, in \u003cem\u003eCrime and Justice in America, 1975-2025\u003c/em\u003e, Chicago, University of Chicago Press\u003c/li\u003e\n\u003cli\u003eSherman, L. W., Gartin, P. R., and Buerger, M. E. 1989. Hot spots of predatory crime: routine activities and the criminology of place, \u003cem\u003eCriminology\u003c/em\u003e, vol. 27, no. 1, 27\u0026ndash;55\u003c/li\u003e\n\u003cli\u003eSherman, L., Neyroud, P. W., and Neyroud, E. 2016. The Cambridge Crime Harm Index: Measuring Total Harm from Crime Based on Sentencing Guidelines, \u003cem\u003ePolicing: A Journal of Policy and Practice\u003c/em\u003e, vol. 10, no. 3, 171\u0026ndash;83\u003c/li\u003e\n\u003cli\u003eSimmonds, D., Ariel, B., and Harinam, V. 2023. Overcoming unreported violence using place‐based ambulance data: The case for mapping hotspots based on health data for crime prevention initiatives, \u003cem\u003eTransactions in GIS\u003c/em\u003e, vol. 27, no. 7, 1928\u0026ndash;41\u003c/li\u003e\n\u003cli\u003eStatistics, O. for N. 2023. \u003cem\u003eCrime in England and Wales: year ending June 2023\u003c/em\u003e, https://ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/bulletins/crimeinenglandandwales/yearendingjune2023\u003c/li\u003e\n\u003cli\u003eSutherland, A., Strang, L., Stepanek, M., Giacomantonio, C., Boyle, A., and Strang, H. 2021. Tracking Violent Crime with Ambulance Data: How Much Crime Goes Uncounted?, \u003cem\u003eCambridge Journal of Evidence-Based Policing\u003c/em\u003e, vol. 5, nos. 1\u0026ndash;2, 20\u0026ndash;39\u003c/li\u003e\n\u003cli\u003eThaler, R. H. and Sunstein, C. R. 2009. \u003cem\u003eNudge: improving decisions about health, wealth and happiness\u003c/em\u003e, London, UK, Penguin\u003c/li\u003e\n\u003cli\u003eWeinborn, C., Ariel, B., Sherman, L. W., and Dwyer, E. O. 2017. Hotspots vs. harmspots: Shifting the focus from counts to harm in the criminology of place, \u003cem\u003eApplied Geography\u003c/em\u003e, vol. 86, 226\u0026ndash;44\u003c/li\u003e\n\u003cli\u003eWeisburd, D. 2015. The law of crime concentration and the criminology of place, \u003cem\u003eCriminology\u003c/em\u003e, vol. 53, no. 2, 133\u0026ndash;57\u003c/li\u003e\n\u003cli\u003eWeisburd, D., Groff, E. R., and Yang, S.-M. 2012. The criminology of place: Street segments and our understanding of the crime problem, in Oxford, UK, Oxford University Press\u003c/li\u003e\n\u003cli\u003eWilliams, S. A. 2015. \u0026ldquo;Do Visits or Time Spent in Hot Spots Patrol Matter Most? A Randomised Control Trial in the West Midlands Police\u0026rdquo;\u003c/li\u003e\n\u003cli\u003eWilliams, S. and Coupe, T. 2017. Frequency Vs. Length of Hot Spots Patrols: a Randomised Controlled Trial, \u003cem\u003eCambridge Journal of Evidence-Based Policing\u003c/em\u003e, vol. 1, no. 1, 5\u0026ndash;21\u003c/li\u003e\n\u003cli\u003eZimring, F. E. and Hawkins, G. J. 1973. \u003cem\u003eDeterrence: The Legal Threat in Crime Control\u003c/em\u003e, Chicago, IL, University of Chicago Press\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 6 are available in the Supplementary Files section\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"cambridge-journal-of-evidence-based-policing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cjep","sideBox":"Learn more about [Cambridge Journal of Evidence-Based Policing](http://link.springer.com/journal/41887)","snPcode":"41887","submissionUrl":"https://submission.nature.com/new-submission/41887/3","title":"Cambridge Journal of Evidence-Based Policing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Hotspots Policing, Prevention of Crime, Crime Reduction, Randomised Controlled Trial, Violent Crime, Application-based Tasking","lastPublishedDoi":"10.21203/rs.3.rs-4650164/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4650164/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e \u003cp\u003eTest the impact of mobile application-targeted patrols of Joint Operations Unit officers at hotspots of violent crime on levels of patrol conducted, and levels of violence in those locations.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eCrossover-randomized experiment, with hotspots (n\u0026thinsp;=\u0026thinsp;45) randomly allocated using a mobile phone-based tasking application, with half allocated each day. Impact on levels of patrol and of violent crime were examined using t and chi-squared tests.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eTasking via app led to dramatically increased officer time in hotspots. An 8.74% decrease in violent crime was seen but was non-significant, and with lower effect sizes than have been found elsewhere.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eUsing a tasking application provided a cost-effective mechanism for achieving hotspots patrols using business-as-usual resources. Traditionally designed hotspots didn\u0026rsquo;t appear optimal for policing violent crime in Thames Valley, and these need to be redesigned to work for the geography and crime density in Thames Valley. Further research should test redesigned hotspots.\u003c/p\u003e","manuscriptTitle":"Testing Application-Based Tasking and Hotspots Policing: A Two-in-One Randomised Trial","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-09-12 12:10:08","doi":"10.21203/rs.3.rs-4650164/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-11-17T19:40:43+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-11-05T13:40:09+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"69662402078764442069292797954674148310","date":"2024-10-19T11:07:17+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-09-23T15:35:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"94366545621552701908097797174570130143","date":"2024-09-18T10:32:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"138496843543080084926305223834329607783","date":"2024-09-16T05:21:24+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-09-15T07:27:56+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-09-08T18:26:00+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-08-14T23:18:03+00:00","index":"","fulltext":""},{"type":"submitted","content":"Cambridge Journal of Evidence-Based Policing","date":"2024-06-27T16:24:45+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"cambridge-journal-of-evidence-based-policing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cjep","sideBox":"Learn more about [Cambridge Journal of Evidence-Based Policing](http://link.springer.com/journal/41887)","snPcode":"41887","submissionUrl":"https://submission.nature.com/new-submission/41887/3","title":"Cambridge Journal of Evidence-Based Policing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"6fd92b98-182a-4b42-91c2-e964c7c53bbd","owner":[],"postedDate":"September 12th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-02-17T16:03:03+00:00","versionOfRecord":{"articleIdentity":"rs-4650164","link":"https://doi.org/10.1007/s41887-024-00096-7","journal":{"identity":"cambridge-journal-of-evidence-based-policing","isVorOnly":false,"title":"Cambridge Journal of Evidence-Based Policing"},"publishedOn":"2025-02-11 15:57:40","publishedOnDateReadable":"February 11th, 2025"},"versionCreatedAt":"2024-09-12 12:10:08","video":"","vorDoi":"10.1007/s41887-024-00096-7","vorDoiUrl":"https://doi.org/10.1007/s41887-024-00096-7","workflowStages":[]},"version":"v1","identity":"rs-4650164","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4650164","identity":"rs-4650164","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}