Search this site

The Society of Evidence Based Policing was pleased to contribute to the development of new guidance from the Youth Endowment Fund - Policing, Children and Violence. This guidance is a strong example of what we exist to promote: policing practice that is genuinely shaped by evidence. The YEF's guidance sets out six recommendations for police forces across England and Wales to help prevent children's involvement in violence. The recommendations draw on new YEF research and the best available international evidence, and have been shaped by an expert panel of police officers, commissioners, academics and other specialists. Bringing together evidence and frontline expertise has helped ensure the recommendations are practical, relevant and useful to police forces. The six recommendations are organised around three themes: Treat children as children: Translate child-centred policing into everyday practice. Audit how you safeguard children to prevent violence. Focus on what works: 3. Take a problem-oriented policing approach. 4. Stop harmful approaches and deprioritise approaches without a robust evidence base. 5. Improve the quality of stop and search. Do diversion well: 6. Build an effective and fair out-of-court resolution approach. Download the guidance Alongside the guidance, the YEF has also published four new evidence reviews exploring key areas of policing practice: Safeguarding children involved in violence: Insights for 'best practice' in policing Child-centred policing in practice: Insights from children and young people Policing's Response to Children's Exploitation in Drug Markets Review of evidence-based practice in policing to prevent children from being involved in violence You can learn more about the YEF's new guidance at the National Evidence-Based Policing Conference 2026, taking place in Milton Keynes on 23 and 24 September 2026. The YEF will have a stand in the the Exhibition Area, where you can meet the team and find out more about the guidance. The YEF will also be presenting in the breakout sessions on Wednesday afternoon. Make sure you check it out!

August has been quieter than July on the research front - three studies for you for this month, from digital coercive control to complaints against specialist firearms and counter-terrorism officers, plus a look at strip-searching in custody. As ever, if you spotted something I missed, let me know and I'll add it to the next edition. TL;DR? Policing digitally enabled coercive control — police often miss digital coercive control because they focus on single incidents rather than patterns, and officer confidence in spotting it is low across the ranks. Legitimacy under two roofs — a comparison of complaints against Met counter-terrorism and firearms officers finds similar patterns of officer-level complaint concentration, but the two datasets cover different years, so read the comparison with care. Strip searches in police custody — in one English force, Black detainees had consistently higher odds of being strip-searched than White detainees, even after accounting for offence type and vulnerability. Policing digitally enabled coercive control: challenges in responding to and investigating digital abuse Authors: Huber and colleagues Journal: Policing and Society Study design: Mixed-methods study combining police incident data from two UK forces with interviews of officers, third-sector practitioners and victim-survivors Setting: United Kingdom Summary: Digital technology increasingly features in coercive control, but this study found police frequently misidentify it, focusing on isolated incidents rather than the cumulative pattern of monitoring and restriction that defines the offence. Officer understanding and confidence in gathering digital evidence was limited across all ranks, though some forces are building specialist digital media investigator roles. You'll be interested if: you lead domestic abuse investigation, digital forensics, or training and professional development for frontline and specialist officers. Legitimacy under two roofs: citizen complaints against counter-terrorism and firearms command policing compared Authors: Ford and Benning Journal: CrimRxiv (preprint) Study design: Comparative quantitative analysis of complaints records from two Metropolitan Police Service specialist commands Setting: United Kingdom Summary: Comparing complaints against the Counter-Terrorism Command and Specialist Firearms Command, this preprint finds broadly similar patterns — unlawfulness as the leading complaint category, and complaints concentrated among a small number of officers in both commands — alongside a notable difference in appeal rates. The two datasets cover different years, which the authors flag as a limit on direct comparison. You'll be interested if: you work in professional standards, complaints handling, or governance of specialist armed or counter-terrorism policing. Strip searches in police custody and their association with ethnicity/race: evidence from an English force Authors: Ali and Dymond Journal: Criminology & Criminal Justice Study design: Quantitative analysis of custody records (over 25,000 detainees) from one English force Setting: United Kingdom Summary: Black detainees faced consistently higher odds of being strip-searched than White detainees, even after accounting for offence type, use of force, and disclosed vulnerabilities such as mental health or self-harm history. The disparity was more pronounced for Black men and Black children specifically than would be expected from race alone. You'll be interested if: you work in custody, professional standards, or oversight of PACE search powers. And that's August 2026!

By Dr. Ben Havers - Strategic Research Manager, British Transport Police When people hear the term evidence-based policing, they often think about frontline tactics, crime reduction interventions or operational problem-solving. Those things matter, of course, but I sometimes think we undersell one of the biggest contributions evidence can make: helping forces to deliver their strategic ambitions by making more informed decisions - what to prioritise, where to invest, what to stop doing and how to prepare for what is coming next. Most forces are good at responding to today's challenges. The harder task is preparing for the issues we have not fully encountered yet. In a rail environment, those questions can feel especially practical: what capabilities will we need in five or ten years' time? Where will future demand come from? What impact will new technologies have on crime, investigations and public expectations? How might a new station, arena, rail route or major event change local demand? Which emerging risks deserve attention now, before they become operational problems? These are exactly the kinds of questions that evidence-based policing can help answer. 1) Seeing Around Corners One of the most valuable things a force can do is lift its head above the day-to-day pressures and look at what's changing in the wider environment. That is particularly important on the railway, where changes to infrastructure, passenger behaviour, technology and the wider transport system can all affect future demand. That's where horizon scanning comes in. At BTP, we've spent a lot of time considering signals that could affect future policing demand on the railway. Some of these are obvious, such as major rail infrastructure projects, station redevelopments and large-scale events. Others are less obvious, including demographic change, emerging technologies, evolving criminal methodologies and new forms of vulnerability. The reality is that identifying a signal is only the beginning. Knowing that a major infrastructure project such as HS2 is being delivered tells us very little on its own. The useful questions come afterwards. How might it affect passenger volumes, crime patterns, crowding, trespass risk or demand at particular locations? What new partnerships might be needed with rail operators, local authorities, event organisers or other police forces? What skills, training or specialist capability might officers and staff need? This is where research and analysis become important. Evidence-based policing provides a structured way of moving beyond speculation and developing a clearer understanding of what a trend or development might actually mean in practice. We can draw on existing research, analyse internal data, examine comparable examples from other places and commission new studies where significant evidence gaps exist. The same applies to emerging technologies such as artificial intelligence. Most of us can see that AI is likely to have implications for policing, but evidence is needed to understand what those implications actually are. Research can help us assess where AI might support decision-making, where it could introduce risk, what safeguards are needed and whether a tool is improving practice or simply adding another layer of complexity. In short, horizon scanning helps us spot potential challenges and opportunities. Evidence helps us work out what they mean, how seriously to take them and what we might need to do next. 2) Finding Out What Actually Works Evidence-based policing is not just about answering interesting questions. It is about answering the questions that matter most to the organisation. Every force has far more potential research topics than it has time or resources to pursue. Without some way of prioritising activity, research can become fragmented and disconnected from organisational need. That was one of the drivers behind the development of BTP's Areas of Research Interest (ARIs). The objective was not simply to create a list of topics for academics. It was to identify areas where stronger evidence could support better decision-making, guide external research proposals and help focus limited organisational capacity on the questions most likely to add value. The agenda is directly linked to our force Strategy, with 3 ARIs per Strategic Objective. For example, if the force wants to improve outcomes around vulnerability, violence against women and girls, victim care, offender management or emerging technology-enabled threats, evidence gives us a practical route in. Research can help define the problem, identify who is most affected, understand what is already working, test assumptions and explore which approaches are most likely to improve outcomes. A research agenda also helps with practical decisions. It can shape which academic partnerships we pursue, which proposals we support, which topics we invite people to work on and where we need to build a stronger evidence base ourselves. In that sense, evidence-based policing acts as a bridge between strategy and delivery. It turns broad priorities into questions that can actually be investigated. 3) Knowing Whether Change Is Actually Working Perhaps the strongest strategic argument for evidence-based policing is its role in helping organisations understand whether change is actually delivering the benefits that were intended. Policing is constantly investing in new technologies, programmes, processes and ways of working. There is often understandable enthusiasm about the potential of these initiatives, but good intentions and promising ideas do not automatically translate into positive outcomes. Without evaluation, it can be surprisingly difficult to know whether an intervention is genuinely effective, whether it represents value for money, or whether there are unintended consequences that need to be addressed. Research and evaluation provide the mechanism for answering those questions. They help organisations move beyond anecdote and assess what is actually happening in practice. That might mean looking at whether a pilot is improving outcomes for victims, whether a new process is reducing demand, whether training is changing behaviour, or whether a digital tool is helping officers make better decisions. Just as importantly, evaluation gives leaders evidence on which to base decisions about whether an initiative should be expanded, adapted or discontinued. A useful example is BTP's work with the College of Policing to evaluate an artificial intelligence tool designed to support stop and search supervision. The final results are awaiting publication, but the project is a good example of the value of testing promising technology properly before making decisions about wider adoption. It has helped ensure that any future decisions are informed by evidence about the tool's benefits, limitations and practical use in a policing context. For me, that is exactly what good evaluation should do. It creates a structured way to test promising ideas, understand how they work in practice and consider any benefits, limitations or risks before decisions are made about wider use. In areas such as AI, where the pace of change is fast and the consequences of getting it wrong can be significant, that kind of careful, evidence-informed approach is especially important. This example captures an important aspect of evidence-based policing that is sometimes overlooked. The purpose of research is not to prove that something works. It is to find out whether it works, for whom, in what circumstances and with what risks or trade-offs. Those are very different things. Evidence that a project has not yet demonstrated sufficient benefit can be just as valuable as evidence that it has. Both outcomes reduce uncertainty, improve future decision-making and contribute to organisational learning. In an environment where resources are limited and demands continue to grow, that learning has real value. So What Does This Mean For Strategic Leaders? Research cannot predict the future with complete accuracy and evaluation will never answer every question. Strategic decisions will always involve judgement. What evidence-based policing can do, however, is improve the quality of that judgement. In practical terms, it can help a force decide whether a promising pilot is ready to scale, whether a new technology is worth adopting, whether a partnership is delivering value, whether a rail or infrastructure development requires investment now, or whether a long-standing way of working needs to change. For me, that is why evidence-based policing should be viewed as much more than an operational methodology. It is a strategic capability. It helps organisations anticipate change, focus attention, test assumptions and learn from experience. At a time when policing faces growing complexity, increasing demand and difficult choices about where to invest resources, that feels less like an academic luxury and more like a core part of good strategic decision-making. Declaration of interest: I work for British Transport Police as Strategic Research Manager and lead work relating to the force's research agenda and research governance. Generative AI was used to support the drafting and editing of this article. The ideas, opinions and conclusions remain my own.

One of the most common phrases heard in policing is, "There must be research on this somewhere." Closely followed by, "If there isn't, someone should look into it." The reality is that both statements are often true. Police officers and staff make thousands of decisions every day. Many are supported by well-established evidence. Others rely on professional judgement because the evidence simply doesn't exist, isn't easy to find, or hasn't yet reached the people who need it most. Our project set out to explore a simple question: What if there was one place where policing could search existing evidence, identify knowledge gaps, and submit ideas for future research? The result was the concept of an Evidence-Based Policing Research Platform—or, as we came to call it, the Ideas Hopper. Start with what we already know One of the core principles of evidence-based policing is making the best use of existing research before creating something new, so we built ChatEBP - an AI-powered chatbot designed to make evidence-based policing quicker and easier to access. Instead of searching through lengthy reports or multiple documents, users can ask questions in plain language and receive clear, evidence-informed answers drawn from a trusted collection of research and guidance. Because ChatEBP only uses carefully selected and approved sources, users can be confident that the information it provides is relevant, consistent, and linked back to the original evidence. This makes it easier to explore research, save time, and build confidence in using evidence to support day-to-day decisions. While ChatEBP has been developed for evidence-based policing, the same approach can be adapted for many other organisations and knowledge areas where trusted, well-managed information is important. The ideas that we never write down A key objective of the Ideas Hopper is to provide a structured way of capturing research questions and evidence gaps identified through policing practice. Where a practitioner identifies an area that could benefit from further exploration, they can submit their idea through the platform. Rather than remaining an isolated suggestion, each submission enters a collaborative review process involving the Society of Evidence-Based Policing, The Open University, and CGI. Drawing on their respective expertise, the three partners assess each submission, considering its relevance to current policing priorities, its feasibility, and its potential to contribute to the evidence base. Through this collaborative assessment process, selected research ideas are identified for further investigation. The Open University then leads the research, with the findings and outputs fed back into the platform, creating a continuous feedback loop that enables new evidence to inform policing practice and inspire further research. Practitioners have visibility of how their ideas are reviewed, developed, and progressed throughout their journey. By bringing together operational insight, research expertise, and digital innovation, the Ideas Hopper provides a transparent and collaborative process for identifying, developing, and translating practitioner-led ideas into evidence-based policing research, ensuring that new knowledge is continually shared back with the policing community. Bringing the idea to life Rather than simply describing our concept, we wanted people to experience how it could work. Working collaboratively across CGI, SEBP and The Open University, we designed an interactive prototype of the Ideas Hopper, allowing users to explore how they could search for evidence, submit research ideas and follow projects as they progressed through the research pipeline. Alongside the prototype, we also created an animated walkthrough that tells the story from a user's perspective. Following the journey of a practitioner, the animation demonstrates how someone can move from identifying an operational challenge to exploring existing evidence, submitting a new research idea and contributing to future policing knowledge. The takeaway Evidence-based policing is built on asking the right questions, sharing knowledge and continually improving practice. While not every operational challenge will have an immediate answer, every question has the potential to contribute to future learning. The Ideas Hopper has been designed to support that journey by creating a collaborative space where practitioners can explore existing evidence, identify gaps and contribute ideas that may shape future research. Through the partnership between the Society of Evidence-Based Policing, The Open University and CGI, the ambition is to make it easier for good ideas to be captured, reviewed and, where appropriate, developed into evidence that benefits policing. After all, today's unanswered question could become tomorrow's evidence. Mia Tsigarides CGI Partner mia.tsigarides@cgi.com

July has been busy! This month’s research runs from money mules and neighbourhood policing to school holidays, AI safeguards, family liaison officers and what the public looks for in a chief. There are 22 papers in the reading pile, including one returning visitor from June. As ever, we have separated what a study found from what its design allows us to conclude. A randomised trial can support a different kind of claim from a survey or interview study; that does not make the latter unimportant. It simply tells us whether we are looking at an effect, an association, an experience or a well-informed explanation-and what question should come next. To make the brief easier to navigate, the studies are grouped under the nearest NPCC committee portfolio. TL;DR Index Crime-linkage models Cybersecurity training formats Domestic-abuse victim satisfaction Family liaison officer wellbeing Focused foot patrols Hybrid recruit learning and attrition Mobility and crime Money muling Neighbourhood policing Outsourced 999 services Perpetrator programmes and criminal thinking Police demand predictors Police leadership quality Police responses to mental-health crises Prison therapeutic communities and reoffending Probabilistic AI in policing and criminal justice SCAN and statement credibility School holidays and crime Sexual-assault police interviews Shift timing and police discretion VR training to prevent ethnic profiling Weapon-ban zones and crime displacement Reading note: ‘How confident can we be?’ is about the claim supported by the design, not a league table of good and bad research. Percentages are rounded where that makes the practical message easier to read. Crime Offending patterns and the quality of police response to people affected by crime. Linking serial crimes: accuracy is only half the operational question Setting: International evidence from 10 countries (most studies were from the UK) 🇬🇧 🇫🇮 🇺🇸 🇨🇦 🇯🇵 🇧🇪 🇳🇱 🇮🇹 🇦🇺 🇿🇦 Source: The Accuracy of Statistical Models to Link Serial Crimes: A Meta-Analysis Who did it? Gauri Milind Kelkar, Christopher Jones and Jessica Woodhams. What was the study about? This was a meta-analysis: a study that systematically brings together results from several earlier studies so the overall pattern can be estimated. It included 29 studies and 206 model results, covering research published to the end of 2022. Most studies concerned burglary or sexual assault and tested whether geography, modus operandi or combinations of behaviour could distinguish linked from unlinked offences. What did it find? Geographical information and combined measures of modus operandi tended to perform best, although the strongest input varied by offence. Reported discrimination was often high. For example, geographical models for burglary produced an average ‘area under the curve’ of about .88. This score describes how well a model separates linked from unlinked cases: .50 is no better than chance and 1.00 is perfect separation. Even a high score did not always translate into a manageable number of false leads. How confident can we be? Combining studies gives a valuable overview, but 79% of the included studies were judged at high risk of selection bias—meaning the cases included may have been systematically easier or different from the cases encountered in normal practice. Many used solved series or samples containing only known serial offenders, which can make performance look better than it would in live, unsolved data. Some offence types were represented by very few studies, and a single accuracy statistic can conceal how many candidate links an analyst must examine. What does it mean in practice? Use models to prioritise human review, not to declare that crimes are linked. Before deployment, test on local unsolved cases and report operational measures: how many genuine links appear near the top, how many false alarms are produced and what workload follows. You’ll be interested if… you work in crime linkage, behavioural investigation, serious sexual offences, burglary analysis or evaluate analytical software. Domestic abuse: feeling heard was closely tied to satisfaction Setting: United Kingdom 🇬🇧 Source: Understanding Victim Satisfaction: Police Responses to Domestic Abuse in Wales, UK Who did it? Daniel Welch. What was the study about? Thirty-eight people with recent police contact about domestic abuse completed an online survey distributed through specialist charities in South Wales, Gwent and Dyfed-Powys. The 89-item survey combined rating scales with open comments. The paper was published in 2026 but does not state the fieldwork dates. What did it find? In this sample, satisfaction was positively associated with officers being approachable, listening and providing support. Feeling that officers were dismissive or judgemental had the strongest negative relationship with satisfaction. Participants described both positive and negative experiences; respect, empathy, validation, clarity and practical help featured repeatedly in how they judged the response. How confident can we be? The mixed-method design gives direct insight into recent lived experience, but the sample is very small, self-selecting and recruited through support organisations. It did not include North Wales and was predominantly White British women. Correlations do not show that any single behaviour caused satisfaction, willingness to proceed or a case outcome. What does it mean in practice? The first response is an operational intervention in its own right. Listening, respectful communication, clear explanations, validation and practical support are concrete behaviours that can be trained, supervised and reviewed. Feedback and body-worn video can help teams examine how those behaviours appear in practice. You’ll be interested if… you lead domestic-abuse response, victim services, first-response training, supervision or work on victim engagement and withdrawal. Criminal Justice Interviewing, evidence and interventions across the criminal justice process. SCAN did not reliably separate truthful from fabricated accounts Setting: Slovenia 🇸🇮 Source: Inefficacy of the SCAN Technique in Discriminating Between Truthful and Fabricated Statements Who did it? Igor Areh and Marija Umolac. What was the study about? In Slovenia, 307 university students wrote either a truthful or fabricated account of a recent negative event. Four independent raters, blind to the truth, received ten hours of training and coded the statements using 11 Scientific Content Analysis-or SCAN-criteria. The paper was published in 2026; it does not report when data were collected. What did it find? Agreement between raters was generally low. The total SCAN score did not differ between truthful and fabricated statements. Two individual criteria differed, but they were not coded reliably. Raters classified 71% of truthful accounts and 58% of fabricated accounts correctly; the latter is close to chance and the pattern may reflect a tendency to label statements truthful. How confident can we be? For the narrow question tested, known ground truth, blind independent raters and a reasonably large sample are useful strengths. The exercise involved students, written statements and fabricated personal events rather than suspects and real investigations. It may not reproduce every way SCAN is taught or used. Even so, the result aligns with earlier independent research and gives no support for using the technique as a truth test. What does it mean in practice? Do not use SCAN scores to decide whether a person is truthful, to close an enquiry or to direct intrusive action. Investigative decisions should rest on evidence-preserving interviews, corroboration and methods with demonstrated reliability. You’ll be interested if… you work in investigative interviewing, intelligence assessment, training, professional practice or assurance of legacy techniques. Sexual-assault interviews: what victim-survivors say would help Setting: Australia 🇦🇺 Source: Sexual Assault Victim-Survivors’ Perceptions of Police Interview Techniques: Results from a Survey Study Who did it? Gemma Hamilton and Dirkje Gerryts. What was the study about? An Australian online survey ran from June to August 2025 and included 285 adult victim-survivors aged 18–71; 72% had reported their experience to police. Participants rated the importance or comfort of interview settings, support, communication and recognised interviewing techniques. The questions captured perceptions, not necessarily personal experience of every technique. What did it find? A private, comfortable setting and a sense that the interviewer believed them were rated most important. Access to support, clear expectations, free narrative and accessibility were also highly valued. Participants were most comfortable being able to correct the interviewer and receiving regular emotional check-ins. Silence and long pauses, interviewer self-disclosure and context reinstatement were less comfortable on average. How confident can we be? This is valuable direct input from victim-survivors across Australia. It does not test which technique produces the most complete or accurate evidence, or whether it improves case outcomes. Recruitment was self-selecting, the sample was predominantly cisgender women, and survey ratings cannot recreate the pressure of a real interview. What does it mean in practice? Interview quality includes preparation and explanation as well as technique. Offer privacy, breaks and support; explain why pauses or memory prompts may be used; check understanding and emotional state; and make it safe to correct the interviewer. Preferences should inform-not replace-the evidential purpose of the interview. You’ll be interested if… you interview victim-survivors, design sexual-offence services, train interviewers or review trauma-informed practice. Therapeutic communities in Illinois prisons: promising, not final Setting: United States 🇺🇸 Source: The impact of therapeutic communities on recidivism in Illinois Who did it? Damon M. Petrich, David E. Olson and Pamela Carey. What was the study about? The researchers studied men released from Illinois prisons in 2018 and 2019. They compared participants from two prison-based therapeutic communities for substance-use disorder with similar men released from other prisons, using matching to balance observed characteristics. Arrest and reincarceration were measured at one and three years. What did it find? Participation in both therapeutic communities was associated with lower general arrest and reincarceration at one and three years. The medium-security programme was also associated with lower violent, drug and domestic-abuse arrests. Differences reduced over time, suggesting that any benefit may weaken after release. How confident can we be? Matching creates a more credible comparison than raw outcome rates, but people were not randomly allocated. Differences that were not measured-motivation, selection into the programme, prison environment or aftercare-may explain part of the result. The study covers men in one US state and compares programmes with ordinary imprisonment rather than a waiting list. What does it mean in practice? Substance-use treatment in custody may support safer re-entry, but the pathway after release deserves equal attention. Track outcomes over several years, distinguish offence types and test how programme fidelity, security setting and community aftercare shape results. You’ll be interested if… you work in prisons, probation, integrated offender management, substance-use treatment, re-entry or reducing repeat offending. Perpetrator programme: thinking changed during participation Setting: United States 🇺🇸 Source: Evaluating cognitive change in a batterer intervention program: a prospective examination of criminal thinking Who did it? Monica Solinas-Saunders and Drew Saunders. What was the study about? The study followed 624 court-mandated participants in a cognitive-behavioural batterer intervention programme across five counties in one Midwestern US state. Participants completed repeated measures of criminal thinking during the programme and for up to two years, producing 1,717 observations. What did it find? Criminal-thinking scores declined during participation and then appeared to stabilise rather than rebound after the programme. The study did not measure a corresponding change in domestic-abuse behaviour or reoffending. How confident can we be? Repeated measures provide useful evidence of within-person cognitive change. Without a randomised or matched comparison group, the study cannot show that the programme caused the decline. Measures were self-reported, the sample was mostly male and White, and longer-term follow-up was very sparse-only six people contributed data at one year after completion. What does it mean in practice? A standardised measure of thinking can be one indicator of progress, but it is not a proxy for victim safety or repeat offending. Programmes should combine proximal measures with completion, risk, police and victim outcomes, preserve fidelity and plan follow-up before participants disappear from the dataset. You’ll be interested if… you commission perpetrator programmes, manage court mandates, assess domestic-abuse interventions or need measures of change before reoffending data mature. Digital, data and technology How AI and data-driven tools are governed, evaluated and used in practice. Probabilistic AI: useful tools, unfinished safeguards Setting: United Kingdom 🇬🇧 Source: The Use of Probabilistic AI in Law Enforcement and the Criminal Justice System of England and Wales Who did it? Temitope Lawal, Kyriakos N. Kotsoglou, Evdoxia Taka, Marion Oswald, Michele Sevegnani and Muffy Calder, for PROBabLE Futures. What was the study about? The team interviewed 25 people from policing, prosecution, the Bar, government, oversight, research and technology providers between May 2025 and January 2026. It combined thematic analysis with a review of policies and system documentation to map how probabilistic AI is entering law enforcement and criminal justice in England and Wales. What did it find? Participants described real value from carefully targeted tools, including transcription, redaction, vulnerability identification, crime linkage, accessibility and officer welfare. They also described governance and institutional capability developing more slowly than adoption. Recurring risks included automation bias, loss of professional skill, unclear accountability, poor data, opaque suppliers, weak testing, deepfakes and errors passing from one AI system into the next. The report makes 26 recommendations across government, PoliceAI, the College of Policing, CPS, judiciary, providers and researchers. How confident can we be? This is a detailed qualitative map of a fast-moving system, not an impact evaluation. Its strength is the range and depth of expert perspectives and the practical connection between findings and recommendations. Its purposive sample is not statistically representative, probation was not represented, and interview evidence cannot tell us how often a risk occurs or whether a particular tool improves outcomes. What does it mean in practice? Start with the operational problem and the standard of evidence it demands. Require evaluation under realistic conditions, publish limitations and error rates, record where AI touches evidence, maintain skills that still need human judgement and design review so people genuinely assess an output rather than merely approve it. ‘Human in the loop’ needs a job description. You’ll be interested if… you buy, build, govern or use AI; oversee disclosure and evidence; design workforce training; or need an AI register and evaluation standard. Diversity, Equality and Inclusion Fairness, disproportionality and how training may influence police practice. VR training on ethnic profiling: improvement may begin with discomfort Setting: Netherlands 🇳🇱 Source: It Gets Worse Before It Gets Better: Individual and Group-Level Effects of a VR-Based Training to Prevent Ethnic Profiling in Policing Who did it? Bas Böing, Peter W. De Vries, Stéphanie M. van den Berg and Ellen Giebels. What was the study about? The ten-month Dutch study involved 519 officers at nine police stations. Stations received one VR session, three repeated sessions or no training, with surveys before training, after sessions and six months after the final session. Assignment reflected station characteristics rather than random allocation. What did it find? A single session mainly improved knowledge and psychological safety. Repeated training initially produced less favourable ratings of attitudes and individual and group confidence than one-off training, but it was the repeated condition that showed sustained longer-term gains in attitudes and perceived effectiveness. Repetition did not add a clear psychological-safety benefit. How confident can we be? Repeated measurement and multiple stations strengthen the study. The team also preregistered the research—recording the main questions and analysis plan before seeing the results—and made the data openly available. The stations were purposively rather than randomly assigned, outcomes were self-reported and only 245 participants completed the final survey. The research did not measure actual stops or public experience, so it supports a plausible training-development pattern, not proof of changed policing behaviour. What does it mean in practice? One-off training can raise awareness, but durable development may need repetition, reflection and time. A temporary dip in confidence need not mean failure: it may reflect a more realistic appreciation of the task. Support that phase and pair self-report with stop data, supervision, complaints or observed practice. You’ll be interested if… you commission stop-and-search, disproportionality or immersive training and need to judge both immediate reactions and longer-term change. Finance Commercial models, contracting and accountability in operational services. Outsourcing 999 services: contracts contain discretion too Setting: United Kingdom 🇬🇧 Source: Police discretion, outsourcing, and performance management: 999 emergency calls in the Lincolnshire Police–G4S Strategic Partnership Who did it? Adam White. What was the study about? The article examines Lincolnshire Police’s £229 million strategic partnership with G4S, which ran from April 2012 to March 2022, focusing on 999 calls and the force control room. Its main evidence is 13 interviews with senior police, staff and G4S participants conducted between September 2022 and January 2023, supported by public and internal performance data and earlier interviews. What did it find? Performance targets did not simply control the partnership. Participants described adjusting targets to improve delivery or accommodate demand-‘contractual discretion’-and sometimes needing to redraw the contract when existing terms blocked a workable response-‘non-contractual discretion’. The partnership mixed achievements and problems and ultimately ended with the service returning to police control. How confident can we be? This is a rich case study of a rare long-term arrangement. It is not a comparison showing that outsourcing performs better or worse than in-house delivery. The senior, purposive sample is small and retrospective; interviewees may remember or frame events differently. The conceptual account is useful for understanding governance, but should not be generalised mechanically to other contracts. What does it mean in practice? Write for change as well as for steady state. Contracts for operational services need clear data access, escalation, public-interest safeguards, agreed ways to change targets and credible exit arrangements. Governance should preserve room to respond to demand without making accountability disappear between organisations. You’ll be interested if… you commission control-room or other police services, manage strategic suppliers, set performance regimes or plan insourcing and contract exit. Local Policing Place, visibility, neighbourhood work and community-facing demand. School holidays change the crime mix, not just the crime total Setting: Sweden 🇸🇪 Source: School Breaks and Crime: Quasi-Experimental Evidence from Sweden Who did it? Niklas Jakobsson, Manne Gerell, Per Kristensson and Siri Jakobsson Støre. What was the study about? The researchers used the staggered timing of Sweden’s winter sports break as a natural comparison. They analysed reported crime across all 290 municipalities for 208 weeks from 2021 to 2024, 60,320 municipality-weeks in total and compared places on holiday with places that were not yet on holiday. What did it find? Reported assaults fell by about 13% during the break. The reduction was concentrated among children and young people: about 46% for ages 7–14 and 32% for ages 15–17, with no clear reduction for younger children or adults. Residential burglary rose by about 14%, and crime increased in skiing destinations receiving visitors. The analysis found no clear sign that crime simply moved to the week before or after. How confident can we be? Staggered dates, checks for pre-existing trends and several robustness tests make this stronger than a simple before-and-after comparison. It is still not a randomised experiment. It relies on recorded crime, covers selected offences and cannot rule out every change in reporting or movement between places. The most defensible conclusion is that the break caused a change in the pattern of recorded crime in this setting. What does it mean in practice? School calendars can be treated as a quantifiable operational variable. Youth assault prevention may be most valuable around school-time routines, while burglary prevention and patrol planning can shift towards residential areas and visitor destinations during breaks. You’ll be interested if… you plan seasonal demand, school and youth safety, burglary prevention, neighbourhood patrols or policing in tourist destinations. Focused foot patrol in Chile: a useful lesson in delivery Setting: Chile 🇨🇱 Source: Focused foot patrols in harm-weighted priority areas: evidence from a randomized field experiment in Chile Who did it? Cristóbal Weinborn, David Weisburd, Álvaro Pérez and Barak Ariel. What was the study about? In Santiago, 159 street segments were identified using both crime frequency and harm. Forty-nine were randomly assigned to receive three 15-minute foot-patrol visits a day and 110 to usual policing. The experiment ran from August 2023 to January 2024, following a six-month baseline. What did it find? Assignment to the patrol programme did not produce a statistically clear reduction in overall crime frequency or harm—in other words, the observed difference was too uncertain to rule out chance. Delivery was much lower than planned: officers completed about 30% of scheduled visits, averaging 0.88 visits a day rather than three. A secondary analysis found a reduction in a group of visible street offences—assault, public alcohol or drug consumption and unlicensed street vending—but this was not the experiment’s main outcome. How confident can we be? The researchers assigned street segments by chance, giving the study a strong foundation for comparing the groups. Low delivery, some activity in control locations and operational changes after allocation make the main test harder to interpret. The clearest conclusion is that the programme as delivered did not reduce aggregate crime or harm. The offence-specific result is promising, but exploratory and in need of a better-delivered replication. What does it mean in practice? The amount of patrol given is part of the intervention, not an administrative footnote. Hotspot trials need visit tracking, supervision and a plan for control-area contamination but how much of that is repeatable in day-to-day policing? Reporting both the assignment and the patrol actually delivered gives leaders a fairer account of whether the idea or the implementation fell short. The problem with hotspot policing may not be the idea so much as whether it can be delivered as conceived. You’ll be interested if… you commission hotspot patrol, use harm indexes, run field trials or need to diagnose why an apparently sound tactic did not move the main outcome. Neighbourhood policing: familiar ingredients, uneven recipe Setting: United Kingdom 🇬🇧 Source: ‘No one knows what we do, until we don’t do it’: defining neighbourhood policing Who did it? Roger Arditti, B. S. Blackburn, Robin Bryant and Hannah Lewis. What was the study about? The researchers interviewed or held focus groups with 28 Metropolitan Police participants between June and September 2021. They included PCSOs, dedicated ward officers, supervisors, senior leaders and chairs of Safer Neighbourhood Boards, and discussed neighbourhood practice during 2020–22. What did it find? Participants broadly agreed that visibility, engagement and problem-solving belong at the centre of neighbourhood policing. They did not share a consistent definition or doctrine for how those ideas should operate. Role ambiguity, abstract expectations, resource pressure, training gaps and uneven delivery recurred, even while participants saw the function as essential to trust and local connection. How confident can we be? The study gathers perspectives across ranks and includes some community representation, giving a useful picture of one large force at one point in time. Participation was purposive and voluntary, the sample was small and mostly white and male, and the study did not test outcomes. It cannot tell us that one definition or operating model produces better trust or crime outcomes. What does it mean in practice? A neighbourhood promise needs an operational minimum: what teams own, what time is protected, how problems are selected, how communities influence priorities and what good work looks like. Local flexibility is valuable, but it works better when it starts from shared foundations. You’ll be interested if… you lead neighbourhood teams, design operating models, protect abstractions from becoming slogans or assess delivery against the neighbourhood policing guarantee. Police as mental-health responders: a system still finding its footing Setting: Canada, Australia, the United Kingdom and Guyana 🇨🇦 🇦🇺 🇬🇧 🇬🇾 Source: De Facto Mental Health Responders: A Contemporary Systematic Literature Review of Police Encounters with Persons Experiencing Mental Illness Who did it? Tyson Alker, Kelly Hine, Nadine McKillop and Tim Prenzler. What was the study about? The systematic review included 20 peer-reviewed empirical studies from Commonwealth countries, focused on contemporary police encounters with people experiencing mental illness. Eight databases were searched using a documented process for finding, screening and assessing studies, and two reviewers checked the evidence. What did it find? Encounters were often crisis-driven and involved people with previous police contact. Arrest, force and involuntary transport remained common possible outcomes. Officers described limited training, unclear guidance and service gaps; people with lived experience described fear, distrust and crisis behaviour being misunderstood. Crisis-intervention and co-response approaches showed some encouraging results, particularly around force, but evidence on arrest and transport was mixed. How confident can we be? The systematic method gives a clear view of the available literature. The underlying evidence is small, heterogeneous and mainly descriptive, with different populations, measures and service systems. It is limited to English-language Commonwealth studies and cannot establish that one response model is best. What does it mean in practice? Police training should help officers recognise how crisis can present, but training alone cannot create a health pathway. Joint protocols, reliable service availability, good handovers and feedback across agencies matter. Evaluate safety, experience, repeat contact and connection to care-not only arrest or transport. You’ll be interested if… you commission mental-health response, co-response, right-care pathways, control-room triage or training for crisis encounters. Operations Operational powers and tactics that may change where activity takes place. Weapon-ban zones: the boundary matters Setting: Germany 🇩🇪 Source: ‘Then they simply go somewhere else’: The effect of hotspot policing on crime development Who did it? Reinhold Melcher, Peer Keßler, Alexandra Heyden, Florian Dittrich and Christoph Meißelbach. What was the study about? The study examined a weapon-ban zone introduced in Leipzig, Germany, in November 2018. It used geocoded police-recorded crime for 24 months before and 24 months after, ending in October 2020, and examined both the zone and surrounding areas using time-series and spatial techniques. What did it find? Violent and property crime fell in the short term inside the zone, with the clearest reduction in theft. The longer-term picture was much less convincing. Crime also appeared to move: the geography and persistence of displacement differed between property and violent offending, with property offending showing the clearer spatial adaptation. How confident can we be? The detailed location data and multiple analyses improve the picture, but this remains an observational before-and-after study. There was no randomised or matched comparison area, no direct measure of police activity, and the period ended during the pandemic. Changes in reporting, enforcement or wider conditions could explain some of the movement. The authors are appropriately cautious: the study cannot prove that the zone caused the changes. What does it mean in practice? A zone should be evaluated outside as well as inside its boundary, over more than its launch period and separately by offence type. If the map stops at the line, displacement can look rather like success. You’ll be interested if… you use exclusion or weapon-control zones, place-based powers, city-centre patrols or hotspot tactics where displacement is a concern. Performance Management Demand, deployment and the context needed to interpret performance fairly. What predicts police demand? A map of the UK evidence Setting: United Kingdom 🇬🇧 Source: The predictors of police demand: a rapid evidence review (Home Office, 2026) Who did it? The Home Office rapid evidence review team. The report acknowledges Ruby Forshaw, Jo Perkins and Abbie Harrison, with external review by Kate Bowers and Tom Kirchmaier. What was the study about? The review searched Google Scholar for UK studies and included 62, mostly published between 2011 and 2021. It coded 1,114 reported links between 49 possible predictors and 27 types of police demand. The search was completed in December 2021, so this is a structured look backwards rather than a live forecast. What did it find? Deprivation, education and income inequality were repeatedly associated with burglary, robbery and theft; unemployment was linked with violence. Population growth, churn and density were generally associated with more crime, while traffic measures were linked with road collisions. Major political or terrorist events were associated with hate crime, and some sporting events with criminal damage, theft and violence. Lockdowns shifted demand rather than simply removing it: burglary and robbery fell, while cybercrime and domestic abuse rose in parts of the evidence. How confident can we be? This is a useful map of a broad and scattered literature, not a test showing that any one factor causes demand. The evidence is uneven across topics, often uses different definitions and is now several years old. Missing people, domestic abuse, mental health and road traffic demand were among the thinner areas. A rapid review using one search platform is also less exhaustive than a full systematic review. What does it mean in practice? Demand planning works best when national patterns are treated as prompts for local testing. Forces can combine demographic, economic, event and mobility information with their own calls, crimes and harm measures, and keep checking whether relationships still hold as conditions change. You’ll be interested if… you plan resources, forecast demand, run a force management statement, commission analytical work or are trying to understand why demand has changed. Mobility and crime: footfall can sharpen the demand picture Setting: England and Wales 🇬🇧 Source: The association between mobility and crime in England and Wales Who did it? Hulya Seyidoglu, Jose Pina-Sánchez, Nick Malleson and Graham Farrell. What was the study about? The researchers linked monthly police-recorded crime with Google Community Mobility data for 328 local authorities in England and Wales from March 2020 to October 2022. The models compared changes within each area over time rather than simply comparing one place with another. What did it find? More time around home was associated with lower property crime, violence and public-order crime. More activity around retail, grocery, transport and workplaces was generally associated with higher crime. The size and shape of the relationship varied by offence and local authority. How confident can we be? The national coverage and repeated observations are useful, and the modelling controls for stable differences between areas. It still shows association, not cause. The pandemic was an unusual period, Google data represent opted-in location users rather than everyone, and monthly averages can hide short-lived or very local patterns. Other changing conditions may account for part of the relationship. What does it mean in practice? Mobility measures can complement calls, crimes, events and local intelligence when forecasting demand. They are best used as a local signal rather than as a national conversion formula: ten points of extra footfall will not mean the same thing everywhere. You’ll be interested if… you forecast demand around transport, retail, nightlife, major disruption or changing patterns of work and travel. The end of a shift changes activity and selectivity Setting: United States 🇺🇸 Source: Tough Shift: The Temporal Dynamics of Police Discretion Who did it? Marc Olson and Ian T. Adams. What was the study about? The study used five years of traffic citation data from Phoenix Police Department: 291,352 citation stops made by 2,924 officers. The researchers inferred ten-hour shifts and compared the same officer at the same time and location when that hour fell at a different point in their shift. What did it find? Citation activity was about 25% lower near the end of a shift than earlier. Officers made fewer citation stops, but recorded more violations per stop, which is more consistent with increased selectivity than with making extra stops to generate overtime. Proactive activity was highest around the middle of the shift. How confident can we be? The large dataset and within-officer comparisons reduce many obvious alternative explanations. Shifts had to be inferred rather than read from rosters, the data contain citation-producing stops rather than all potential encounters, and the study concerns one agency and one shift pattern. Some findings about selectivity became less stable under the strictest comparisons. What does it mean in practice? When leaders interpret activity, shift position belongs in the conversation. Briefings, overlap, handovers, report-writing and recovery can be designed around predictable changes in proactivity. End-of-shift decline should not be treated as evidence of motive without looking at workload and opportunity. You’ll be interested if… you design shifts, supervise proactive policing, interpret officer activity or study fatigue and discretion. Prevention Reducing exposure and helping people act before harm or victimisation occurs. Money muling: exposure is wider than understanding Setting: United Kingdom 🇬🇧 Source: UK public experiences, perceptions and understanding of money muling (Home Office, 2026) Who did it? Amy Scoffham and Jessica Kelly, Home Office, using surveys delivered by Ipsos UK. What was the study about? Two online surveys asked people aged 16–75 about approaches to move money for someone else, their understanding of the law and whether payment would change their willingness. A random-probability panel—where people are selected through a random process with a known chance of selection, rather than simply volunteering—surveyed 2,114 people in Great Britain from 15–21 February 2024. A separate quota sample surveyed 2,233 people across the UK from 22–28 March 2024; this second group was recruited online and balanced to match selected population characteristics, but was not randomly selected. What did it find? Fourteen per cent had seen a general opportunity to become a money mule and 6% had been approached directly. Exposure was higher among younger adults: a quarter of 16–24-year-olds had seen an opportunity advertised. Email featured most in direct approaches, while social media featured most in general advertising. Almost everyone said they would refuse, even as the payment rose, but only 18% correctly recognised all three illegal scenarios presented to them. People who had encountered recruitment were more likely to misclassify at least one scenario. How confident can we be? The random-probability survey gives a useful national baseline because selection does not depend only on who chooses to take part, and the second survey points in much the same direction. These are self-reports, however, and the payment questions are hypothetical. Seeing an approach is not the same as accepting one, and the quota survey does not represent people who are digitally excluded. What does it mean in practice? Prevention messages may need to do more than say ‘money muling is illegal’. They can show the less obvious arrangements that count, the approaches people actually receive and the consequences of allowing an account to be used. Younger adults and the channels on which recruitment occurs are obvious priorities, but the survey also shows this is not solely a youth issue. You’ll be interested if… you work on fraud, economic crime, safeguarding, schools and universities, communications, social-media intelligence or victim prevention. Cybersecurity training: short video beat written advice on intentions Setting: United States🇺🇸 Source: Video-based cybersecurity training and protection motivation: evidence from a randomized experiment Who did it? George W. Burruss, Courtney L. Weber and Taylor Fisher. What was the study about? The researchers randomly assigned 437 adults in the United States to watch a narrative video, watch a direct instructional video or read a text warning about phishing. The online study ran from 10 July to 1 August 2024 and measured intended cyber-hygiene behaviour before and immediately after the material. What did it find? Both videos increased protective intentions more than the written message. There was no meaningful difference between the story-led and instructional videos. Believing that the recommended actions would work, feeling vulnerable, seeing the threat as serious and perceiving fewer practical barriers were linked with stronger intentions. Technical skill and reflective thinking also mattered. How confident can we be? Random assignment supports a causal claim about the immediate change in self-reported intentions in this study. It does not show that participants later behaved differently, retained the message or resisted a real phishing attempt. The text condition was also an intervention, and the analysis of possible psychological mechanisms was correlational rather than causal. What does it mean in practice? Short, clear video may be a useful step up from written advice. The content should emphasise actions that work and make them easy to complete. Before investing heavily in drama versus direct instruction, test both against behaviour-such as a safe simulated phishing exercise-and check whether any gain lasts. You’ll be interested if… you deliver cybercrime prevention, staff security training, public communications or behaviour-change campaigns. Workforce Recruitment, learning, wellbeing and public expectations of leadership. Hybrid recruit learning: attrition had more than one address Setting: Australia 🇦🇺 Source: Is online/hybrid learning for police recruits effective and accessible and impacting attrition rates of police recruits? Who did it? Amanda Davies and Antony Stephenson. What was the study about? The study followed one full-time and one part-time 2025 cohort on the New South Wales Police Force Associate Degree in Policing Practice. More than 80% of the first phase was online. Surveys and interviews explored accessibility, learning and attrition among active and disengaged recruits. What did it find? Active recruits generally felt able to use the technology and study online. The small number of disengaged responses did not suggest that online delivery was the main or sole reason for leaving; family, finance and wider life pressures appeared more important. Part-time recruits often seemed particularly prepared for the format. How confident can we be? This is an early, useful temperature check rather than a firm evaluation. Only two disengaged full-time recruits and three disengaged part-time recruits completed a survey, none took part in interviews, and active recruits self-selected into the research. It covers one academy and two cohorts, so it cannot establish whether delivery mode affects attrition. The work forms part of a longer five-year study. What does it mean in practice? Do not diagnose online learning as the attrition problem without following the people who leave. Track reasons for withdrawal consistently and build support around finance, family demands, workload, digital access and academic confidence. Compare outcomes as well as satisfaction over time. You’ll be interested if… you design recruit education, blended learning, workforce entry routes, retention support or evaluation of the police education model. Family liaison officers: protection is more than personal resilience Setting: United Kingdom 🇬🇧 Source: The Role of Resilience and Internal Locus of Control as Protective Factors Against Secondary Trauma and Poor Wellbeing in UK Police Family Liaison Officers Who did it? Jason Roach, Ashley Cartwright and Liam Scott Curran. What was the study about? An online survey of 394 family liaison officers in England and Wales ran from February to May 2024. It used established scales to measure secondary traumatic stress, wellbeing, resilience and the degree to which people felt able to influence events in their lives. What did it find? Average secondary traumatic stress was in the mild range, but experience varied sharply: 15% scored in the severe range and a further 7% in the high range. Personal distress during the previous year was the strongest predictor of poorer wellbeing and greater secondary trauma. Higher resilience and a stronger internal sense of control were associated with better outcomes. How confident can we be? The study has a substantial specialist sample and uses recognised measures. It is cross-sectional, however: a snapshot taken at one point rather than a study following change over time. It therefore cannot show that resilience or a sense of control causes better wellbeing, or that resilience training would solve the problem. People under greatest strain may have been more or less likely to respond. What does it mean in practice? Offer confidential, whole-person support and regular supervision; monitor exposure, workload and recovery; and give officers meaningful voice and control where the role allows. Individual coping skills may help, but the findings do not justify placing responsibility for occupational harm on the officer alone. You’ll be interested if… you support FLOs or other trauma-exposed roles, commission wellbeing provision, supervise specialist teams or assess occupational health risk. What signals a good chief to the public? Credentials count-but need translation Setting: United States 🇺🇸 Source: What signals police leadership quality to the public? A look at evidence-based policing and professional standards Who did it? Gordon Abner and Emily Mee. What was the study about? In April 2026, 303 adults in the United States completed a randomised choice exercise. They repeatedly chose between fictional police-chief candidates whose evidence-based policing experience, professional standards, experience, current role and population served were varied. A follow-up invited open comments on how people made sense of the credentials. What did it find? Professional standards and evidence-based policing were the two most influential attributes in the choices, closely followed by experience serving a relevant population. External credentials-such as accreditation or an evidence-based policing award-were preferred to descriptions of implementation experience. Comments suggest credentials acted as shortcuts for verification and prestige, although some people questioned their politics or distance from day-to-day practice. How confident can we be? Randomised profiles allow a sound comparison of preferences within the task. They do not show whom people would support in a real appointment, whether a credentialed chief performs better or whether trust changes after appointment. Results depend on the explanations and five attributes supplied, and the US online sample limits direct transfer to UK policing. What does it mean in practice? Independent standards and evidence credentials may reassure the public, but labels should be accompanied by concrete examples: what changed, how it was tested and who benefited. Appointment panels should also avoid treating a credential as a substitute for the competence it is meant to signal. You’ll be interested if… you recruit or develop chief officers, communicate professional standards, lead accreditation or explain evidence-based policing to the public. And that’s July 2026!

The Lincolnshire coast is one of the county’s greatest assets, a vibrant tourism economy, vast natural beauty, and a place of rest for millions each year. Yet beneath this success sits one of the most complex safeguarding environments in the region. This blog sets out how Lincolnshire Police and partners moved beyond reactive responses to domestic abuse and implemented a data‑driven, preventative model that reduced harm, protected children, and embedded safeguarding directly into the holiday environment.   The Lincolnshire coastline spans over 680 square miles, includes 50+ miles of coast, two Sites of Special Scientific Interest, and areas of restricted airspace due to an RAF weapons range.During peak season, a resident population is temporarily overlaid by more than four million visitors, concentrated across: 262 caravan parks 40,000 static caravans Large holiday resorts, short‑stay accommodation, and hotels While tourism underpins the local economy, this seasonal surge fundamentally reshapes demand, risk, and vulnerability across policing, health, emergency services, and the voluntary sector. Domestic abuse (DA) emerged as one of the most significant yet historically under‑examined harms within this transient population.   Domestic Abuse: Hidden in Plain Sight For this programme, domestic abuse was defined as a pattern of controlling, coercive, threatening, degrading or violent behaviour, including sexual violence, perpetrated primarily by partners, ex‑partners, family members, or carers. Historically, domestic abuse along the coast was not absent; it was misunderstood. Transient populations arrive with limited or no local history, making conventional safeguarding tools less effective. The reality was stark: Domestic abuse became one of the highest recorded crime types along the coast during summer Skegness and Ingoldmells consistently recorded the highest levels in Lincolnshire Skegness alone accounted for 6% of all DA recorded county‑wide 33–40% of incidents occurred between June and August 30–34% of incidents occurred within caravan and holiday parks This was not random crime. It was seasonal, situational, and predictable.   What the Data Revealed Above is a map of identified hot spot areas relating to Domestic violence. The larger the spot the higher the demand in that area. These larger spots focus on a selective number of our caravan parks.   Five years of crime and call‑for‑service data revealed clear and repeatable patterns: Later peak times on the coast (22:00–02:00) Greater involvement of current partners and family members Offenders from outside Lincolnshire rising from 24% to 38% during summer (predominantly from Nottingham, Sheffield, Leicester, and Mansfield) Higher levels of assault with injury Children more frequently present and witnessing abuse, especially in holiday parks Risk shifted geographically from town centres during the day to caravan parks at night, intensifying around check‑in days and school holidays.   Understanding the “Why”    This project went beyond counting incidents. It asked why holidays intended for rest were becoming periods of heightened risk. Five interacting drivers emerged: Transition stress Incidents peaked shortly after arrival. The move into “holiday mode”, not the holiday itself, proved most volatile. Extended alcohol availability On‑site entertainment and family events created longer and earlier drinking opportunities. Environmental compression Small living spaces removed physical and emotional escape routes, escalating conflict. Financial and emotional pressure Discounted holidays increased access but also intensified stress where expectations failed. Children present in greater numbers Between May and September, 23% of DA incidents involved children witnessing abuse, with some sites showing an 80% increase. The conclusion was unavoidable: this harm was foreseeable and preventable.   A Strategic Shift: From Reaction to Prevention Historically, responses focused on enforcement after harm occurred. This programme deliberately re‑engineered the model, guided by four ambitions: Reduce harm, not just reports Reach victims privately and early Embed safeguarding into the holiday environment Reduce repeat demand on services   Prepare: Quiet Access to Support A bespoke Lincolnshire Coastal Safety Guide was embedded into accommodation: Universal, non‑stigmatising design Multiple safety topics (fire, beach, crime prevention) Healthy relationships section with discreet QR access to DA and VAWG support Crucially, the guide was also shared pre‑arrival through schools in visitor source areas, shifting prevention upstream, before families reached the harm geography.   Protect & Pursue: Matching Resources to Reality   Policing approaches were realigned to match actual demand: Targeted presence on check‑in days and late‑night peak windows Focus on known hotspots and transit routes Licensing engagement aimed at preventing escalation, not post‑incident response Late‑shift staffing redesigned to cover the true 22:00–04:00 risk window Prevent: Making Abuse a Community Issue Safeguarding messages were embedded across: Family venues, cafés and leisure sites Social media and partner platforms Multi‑agency networks through the Atlantis Alliance One campaign post reached 645 shares, outperforming other high‑profile safety messaging that week clear evidence that silence was being challenged.   What Changed: Measurable Impact Despite increased visitor numbers, participating sites saw: Reductions in DA incidents Lower risk grading Earlier, safer access to support Reduced policing demand Reductions in DA incidents at participating parks: Haven Mablethorpe saw a reduction from 12 incidents in 2024 to 5 incidents in the 2025 season. The Grange holiday park in Mablethorpe saw reductions in 2025 compared to 2024. Haven Skegness had 3 of their Domestic Abuse Stalking and Harassment (DASH) Risk assessments graded as medium in 2024. In 2025, all 9 were graded as standard, showing a reduction in risk as well as volume of incidents. Victims accessed help earlier, discreetly, and safely. Children were indirectly safeguarded, and policing demand decreased where engagement was active. By contrast, comparable non‑participating sites experienced significant increases, effectively forming a natural control group. The control group arose because two of the identified hotspot sites were not willing to take part. The providers declined participation due to corporate branding or perception concerns, as they operate a holiday park group with multiple sites across England. One of these sites was directly comparable to the participating parks, containing onsite facilities such as pools, bars and evening entertainment. This non‑participating site saw an increase in DA crime reports, reinforcing the intervention’s impact: Butlins saw an increase from 14 reports in 2024 up to 25 reports in 2025. A second non‑participating site, which does not have onsite venues but is entirely surrounded by bars and clubs, making it similar in design and demand, also saw an increase in incidents, from 2 in 2024 to 10 in 2025. From 2024 to 2025: Visitor numbers increased DA incidents in caravan parks decreased Overall coastal demand rose slightly suggesting place‑based prevention works where embedded Ruling Out Alternative Explanations: Visitor Numbers To ensure that the observed reduction in reports was not attributable to a decrease in visitor numbers, visitor data from selected holiday parks was reviewed. The data covers the period from 1 May to 30 September (inclusive) and excludes sub‑let units and holiday homeowners. The analysis showed the following: Haven Mablethorpe recorded 96,237 guests in 2024, increasing to 108,500 guests in 2025. Haven Skegness recorded 54,158 guests in 2024, increasing to 66,134 guests in 2025. This demonstrates a substantial increase in visitor numbers across both locations during the 2025 period. Despite this uplift in footfall, a reduction in reports was still observed. Learning, Legacy, and Transferability This model demonstrated that: Safeguarding can be designed into places, not added after harm Transient populations are reachable with the right approach Partnership‑led prevention is cost‑effective and sustainable The approach is transferable to any tourist or transient setting   Conclusion: Building Safety Where People Come to Rest Domestic abuse along the Lincolnshire coast is not an inevitable consequence of tourism. By combining data analysis, partnership engagement, environmental design, and victim-centred access, harm can be prevented at scale. This project shifted domestic abuse from a reactive policing issue to a proactive, preventative, and community-wide safeguarding model. The results demonstrate measurable reductions in harm which is grounded in evidence and learning. Allows earlier access to support, and embedding of safeguarding into the holiday environment, creating safer spaces for families and children. Crucially, it shows that policing and partners can act upstream, changing outcomes for families and children before crisis occurs. This was not just a reduction in incidents. It was a fundamental shift in how safety is built into places where people come to rest, reconnect and be protected. There is no holiday from abuse. But there can be no holiday from safeguarding either. Author: Barnaby Prince, Police Community Support Officer, Lincolnshire Police Dementia Safeguarding Lead & Member of the force Safeguarding Network Police lead for the Atlantis Alliance – Collaborative Coastal Working

Easier Said Than Done, Episode 11 Trauma-informed practice has become one of the most talked-about ideas in modern policing and, according to this month's Easier Said Than Done panel, one of the most misunderstood. Matt was joined by two of the field's leading practitioners for an hour that moved from neuroscience to national policy, and from officer wellbeing to public spending: Dr Jane Pepa, former head teacher, now trauma lead for the Merseyside Violence Reduction Partnership, having trained over 12,000 people (including more than 4,000 police officers) in trauma-informed practice. Justin Srivastava retired police superintendent, former trauma-informed lead for the Lancashire Violence Reduction Network, the NPCC's public health portfolio, and the Global Law Enforcement and Public Health Association- There's no single definition and that's the problem Both panellists opened by admitting the field lacks a shared definition. Jane is currently writing her second research paper specifically on this gap, warning that without a substantial, evidence-based definition, "trauma-informed" risks becoming "wishy-washy" - a label anyone can claim after a half-day session. Justin pointed to the SAMHSA model (used by the Department of Health and Social Care) built around principles like psychological safety, trust, empowerment and cultural awareness but noted that even the word "trauma" carries different meanings across health, education and policing audiences, and that getting the narrative right matters as much as the content. The evidence: this is as much about officers as the public A recurring theme was that trauma-informed practice isn't primarily a public-facing intervention - it starts with the workforce. Dr Pepa's "battery pack" analogy that officers can't run on empty, and can't extend compassion to others from a depleted reserve - anchored a wider point: understanding trauma changes how officers see their own experiences, not just those of victims and suspects. The numbers back this up: The average member of the public experiences 6-8 traumatic incidents in a lifetime; the average police officer encounters 600-800. Around 70% of the population has experienced at least one trauma. People with four or more Adverse Childhood Experiences (ACEs) cost an estimated £6,000 per person per year - roughly £20.5 billion UK-wide and are significantly more likely to experience impaired mobility, chronic pain, anxiety and depression. Those with six or more ACEs die, on average, 20 years earlier than the general population. The wider European cost of ACEs has been estimated at £748 billion. As Srivastava put it: prevention and early intervention through trauma-informed practice speaks to "the whole fundamental reason why we joined policing in the first place." The misuse: when trauma-informed becomes a tick-box Both speakers were candid about how the language gets diluted. Short e-learning modules, one-off two-hour sessions, and "trauma awareness" repackaged as "trauma-informed practice" were named as recurring problems, sometimes leaving staff frightened rather than equipped, or organisations wrongly believing a single course means they've "arrived." Jane's view was blunt: genuine practice requires depth (Merseyside's model runs to nearly eight hours, blended with e-learning and ongoing organisational support), plus a "wraparound" safety net of occupational health, supervision, and follow-up so people aren't left to process difficult material alone. The missed opportunity: no national model, but plenty of resources Despite strong local evidence bases, Merseyside, Lancashire, and VRUs across England, Wales and Scotland, there is currently no single, nationally agreed model or curriculum for trauma-informed policing, and training remains inconsistent between forces. This makes the evidence base harder to pin down centrally, even though practitioners see the impact daily. The panel pointed listeners toward existing resources that don't get enough visibility: the College of Policing's public health approaches hub and landscape reviews, the Youth Endowment Fund toolkit, individual VRU websites (which routinely publish their own evaluations), and a newly released UK & Australia scoping review on trauma-informed policing. The takeaway Trauma-informed practice, done properly, is a whole-system, multi-agency commitment - not a training course to complete once. It has to work top-down (mandated by system leaders) and bottom-up (embedded in day-to-day culture), and it has to look after the people delivering it as much as the people receiving it. Or, as Srivastava summed up the case for taking it seriously: "to deal with violence you have to look at the underlying drivers, and to look at the underlying drivers you cannot ignore trauma." Watch the full session on the SEBP members' hub. Membership is free!

By way of introduction, I am a serving Detective Sergeant in the Metropolitan Police and am currently elected to the National Board of PFEW. I hold portfolios on Leadership, Culture and Workloads. I am also Deputy Secretary of the Police Federation National Detective Forum and a non-executive director at the College of Policing. The following blog concerns my dissertation, Canteens to Keyboards , that I wrote in part fulfilment of my degree in MSc Leadership of Policing at the University of Staffordshire. I am currently studying towards a PhD looking at how fractious leadership has impacted on Police culture. When I set out to write my dissertation, Canteens to Keyboards, I was not simply trying to document a shift in communication practices. I was trying to understand something far more fundamental, what happens to police culture when the spaces that once sustained it quietly disappear, and what fills that void in their absence. After more than two decades in policing, much of it as a detective, supervisor, (and Federation Representative), I have seen this shift unfold in real time. What began as an organisational adjustment, closing canteens, rationalising estates and digitising communication, has become something much more consequential. It has altered how officers relate to one another, how they learn, and increasingly, how they are judged. My dissertation starts from a simple but important premise, that police culture has never been static. Academics such as Reiner (2010) and Loftus (2009) have long described it as complex, adaptive, and often contradictory. It can support resilience and solidarity, while at the same time entrenching unhelpful norms. What I argue, however, is that we have misunderstood the function of the environments in which that culture was historically formed. The canteen, the parade room, even the station bar, were not just social luxuries. They were mechanisms of informal regulation, places where behaviour was tested, challenged, moderated, and, importantly, understood within context. The removal of these spaces, accelerated by austerity measures, estate sell-offs, and a growing emphasis on efficiency has not removed police culture. It has displaced it. The conversations still happen, the humour still exists, the storytelling continues. But now it happens on screens rather than across tables. WhatsApp groups replace canteen tables. Microsoft Teams replaces corridor conversations. The shift is not inherently negative, but it is fundamentally different. What concerns me, and what became increasingly clear through my research, is that the organisational response to this shift has been both reactive and misguided. Rather than recognising digital spaces as the new arenas of cultural formation, leadership has often treated them purely as sites of risk. The response has been to regulate, monitor, and, when necessary, punish. This is where the central argument of my dissertation becomes more pointed. Culture cannot be shaped through punitive process alone. You cannot fix culture with a hammer. There is a growing tendency within policing to believe that if we identify misconduct, discipline it robustly, and publicise the outcome, culture will improve. This assumption is not supported by the evidence. Organisational culture does not respond well to blunt instruments. As Schein (2010) argues, culture is embedded through shared experience, reinforced through relationships, and sustained through meaning-making. Punishment may deter specific behaviours, but it does not build the underlying values or social norms that define culture. In fact, there is a strong argument that over-reliance on punitive approaches can be counterproductive. My own experience, supported by emerging research and high-profile cases such as the Operation Hotton and the Casey Review, suggests that when officers feel constantly scrutinised and at risk of sanction, behaviour does not necessarily improve, it becomes more concealed. Informal conversations do not stop; they move to less visible spaces, or they become more guarded and performative. This undermines one of the key functions of informal culture, the ability to challenge and correct behaviour in real time among peers. My dissertation reframes this issue through the concept of social capital, drawing on Putnam (2000). Traditional policing environments fostered strong bonding social capital, tight, trust-based relationships within teams. These relationships were not always comfortable, but they created conditions where individuals could be held to account informally, often more effectively than through formal processes. The decline of physical spaces has weakened these bonds, while digital communication, although efficient, does not replicate the same depth of connection. This is where leadership becomes critical, and where, in my view, current models are falling short. Much of what I have observed and what participants in my research articulated, is a form of leadership that is heavily transactional. It is focused on compliance, risk management, and performance metrics. It is, in many ways, understandable given the pressures policing faces, public scrutiny, political oversight, and organisational risk. However, this model is insufficient for shaping culture. Relational leadership, by contrast, offers a more credible pathway. This is not about being “soft” or avoiding accountability. It is about understanding that culture is shaped through relationships, trust, and credibility. Leaders who are visible, who understand the realities of frontline work, and who engage authentically with their teams are far more likely to influence behaviour than those who rely solely on policy and discipline. Uhl-Bien et al. (2007) describe this within complexity leadership theory as the distinction between administrative and adaptive leadership. Policing has become heavily weighted towards the former, often at the expense of the latter. The consequence is a form of cultural dissonance. Officers are expected to navigate complex, often traumatic situations with discretion and judgement, yet are managed through systems that prioritise control and standardisation. The informal spaces that once allowed officers to process these experiences collectively have diminished, and the digital spaces that have replaced them are subject to surveillance and retrospective scrutiny. The result is a workforce that is increasingly cautious, sometimes cynical, and often disconnected. There is also a legitimacy dimension to this. Public confidence in policing is not shaped solely by crime rates or response times. It is influenced by perceptions of fairness, integrity, and internal culture. The exposure of inappropriate behaviour through digital communication has undoubtedly damaged trust. However, the response cannot be limited to enforcement. Legitimacy, as Tyler (2006) argues, is built through procedural justice, through fairness, transparency, and respectful treatment. These principles apply internally as much as they do externally. One of the more interesting tensions in my research is the coexistence of nostalgia and pragmatism. Many officers I spoke to expressed a sense of loss regarding traditional forms of interaction. They spoke about the value of the canteen as a place to decompress, to learn, and to build relationships. At the same time, they recognised the benefits of digital communication, its speed, its convenience, its ability to connect dispersed teams. The issue, therefore, is not about reverting to the past, but about understanding what has been lost and how it might be replaced or reimagined. Other sectors offer useful comparisons. Research on workplace design and organisational behaviour consistently highlights the importance of informal interaction in fostering collaboration and innovation (Oldenburg, 1999). The concept of the “third place” spaces that are neither home nor formal workplace has been shown to play a crucial role in building social cohesion. Policing, perhaps inadvertently, has removed many of these spaces without fully considering the consequences. My dissertation does not argue for a wholesale return to canteens and station bars, nor does it suggest that digital communication is inherently problematic. Rather, it calls for a more balanced and thoughtful approach. If digital platforms are now central to police interaction, then they must be understood as cultural spaces, not just operational tools. This requires a shift in how they are governed. Instead of viewing them solely through a disciplinary lens, there is a need to consider how they can be used to support positive cultural development. This is where I see the potential for what I describe as cultural leadership. While not yet fully established within the policing literature, the concept reflects an approach to leadership that prioritises meaning, relationships, and shared values. It recognises that culture is not something that can be imposed from above, but something that emerges from within. Leaders, therefore, have a role in shaping the conditions under which culture develops, rather than attempting to control it directly. To return to the earlier metaphor, culture is not a problem to be smashed into submission. It is something to be carefully sculpted. That requires patience, skill, and, above all, understanding. It requires leaders who are willing to engage with the messy realities of organisational life, rather than retreating into process and policy. It also requires an acceptance that culture change is inherently slow and often uncomfortable. There is, of course, a counter-argument. Some would suggest that in an era of heightened scrutiny, robust disciplinary processes are not only necessary but essential. There is truth in this. Misconduct must be addressed, and public confidence depends on it. However, the risk lies in allowing these processes to become the primary mechanism of cultural change. When discipline becomes the dominant narrative, it can overshadow the more subtle, relational work that is required to build a healthy culture. My own experience, both as a practitioner and as a researcher, leads me to a more cautious conclusion. Policing is at a crossroads. The shift from canteens to keyboards is not reversible, nor should it be. But the way in which we respond to it will shape the future of the organisation. If we continue to rely on punitive, compliance-driven approaches, we risk creating a culture that is risk-averse, fragmented, and ultimately less effective. If, however, we invest in relational leadership, rebuild social capital, and engage more thoughtfully with digital spaces, there is an opportunity to create a culture that is both resilient and adaptive. That, ultimately, is the challenge set out in my dissertation. Not to romanticise the past, nor to resist change, but to understand it and to shape it with care rather than force. Author: Paul Matthews, Detective Sergeant, Metropolitan Police Deputy Secretary of the Police Federation National Detective Forum Non-executive director at the College of Policing References  Loftus, B. (2009) Police Culture in a Changing World. Oxford: Oxford University Press. Oldenburg, R. (1999) The Great Good Place. New York: Marlowe & Company. Putnam, R. (2000) Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster. Reiner, R. (2010) The Politics of the Police. 4th edn. Oxford: Oxford University Press. Schein, E.H. (2010) Organizational Culture and Leadership. 4th edn. San Francisco: Jossey-Bass. Tyler, T.R. (2006) Why People Obey the Law. Princeton: Princeton University Press. Uhl-Bien, M., Marion, R. and McKelvey, B. (2007) ‘Complexity leadership theory: Shifting leadership from the industrial age to the knowledge era’, The Leadership Quarterly, 18(4), pp. 298–318.

It's been a quieter month for brand-new studies, but the half-dozen I've gathered are well worth your time - there's work on AI and body-worn cameras, the fairness of camera versus officer enforcement, stop and search, why people do (and don't) report crime, the shifting geography of vehicle theft, and domestic abuse within the police workforce itself. As always, if you spot something I've missed, let me know and I'll add it to the next edition. TL;DR? Artificial intelligence in police research: a preliminary examination of feasibility and replication An AI tool broadly reproduced a de-escalation training trial's findings from body-worn camera footage, but its hard-to-inspect design warrants caution. Camera or cop: the procedural justice of AI-camera and officer enforcement of mobile phone offending - Drivers saw enforcement by an officer as fairer than an AI detection camera, even for the same offence and outcome. Procedural justice in police stop and search encounters: crime contexts and police versus public assessments - Police and the public can judge the fairness of the very same recorded stop-and-search quite differently. The map and the territory: cognitive thresholds in crime reporting - Willingness to report crime varies systematically by demographics, so recorded crime is a partial, socially filtered map of harm. Measuring macro and micro geospatial changes in New York City's motor vehicle thefts - A new spatial method tracks how the geography of vehicle theft shifted during COVID-19 restrictions. Domestic abuse victimisation in a police workforce: victim experience and the institutional response - A qualitative look at the role conflict facing officers and staff who are themselves victims of domestic abuse (preprint). Artificial intelligence in police research: a preliminary examination of feasibility and replication Authors: McLean, Rojek & Nix Journal of Experimental Criminology Study design: Replication study using AI to re-analyse body-worn camera footage from a randomised controlled trial Setting: United States Summary: Researchers tested whether a commercial AI tool (TrustStat) could reproduce findings from a randomised trial of police de-escalation training that originally relied on painstaking human coding of body-worn camera footage. The AI-generated measures reached the same overall conclusion as the human analysis - trained officers communicated more calmly - though several AI and human measures did not align. The authors urge caution given the tool's proprietary, hard-to-inspect design. You'll be interested if: you're weighing whether AI tools can speed up the analysis of body-worn camera footage, or you commission research and evaluation and want to understand both the promise and the limits of off-the-shelf AI for measuring officer behaviour. Camera or cop: the procedural justice of AI-camera and officer enforcement of mobile phone offending Authors: Truelove, Anderson, Bates & Oviedo-Trespalacios Transportation Research Part F: Psychology and Behaviour Study design: Mixed methods (interviews with 26 officers and a survey of 292 drivers) Setting: Australia Summary: Combining interviews with 26 traffic officers and a survey of 292 drivers, this study compared how people perceive the fairness of mobile-phone-offence enforcement by police officers versus AI detection cameras. Drivers rated officer enforcement as fairer and felt more listened to, respected and treated politely, with greater confidence in the decision. Perceptions of 'voice' and trustworthy motives did not differ between the two methods. You'll be interested if: your force is expanding automated enforcement - speed, mobile-phone or other detection cameras — and you want to understand how removing the human interaction may affect public perceptions of fairness, trust and legitimacy. Procedural justice in police stop and search encounters: crime contexts and police versus public assessments Authors: Tansill & Tankebe Criminology & Criminal Justice Study design: Observational study coding 140 stop-and-search encounters on body-worn video Setting: England Summary: Using systematic analysis of 140 stop-and-search encounters captured on body-worn video in an English force, this study examined the quality of police behaviour — judged against procedural justice principles - and how it varied between higher-crime and lower-crime areas. It also offers a rare direct comparison of how police supervisors and members of the public rated the same recorded encounters, surfacing differences in how each group judges identical behaviour. You'll be interested if: you lead or scrutinise stop and search, run a scrutiny panel, or supervise officers — the study speaks directly to how encounter quality is assessed and how police and public judgements of the same interaction can diverge. The map and the territory: cognitive thresholds in crime reporting Authors: Stubbs Criminology & Criminal Justice Study design: Cross-sectional survey (n=1,948) with clustering analysis Setting: England and Wales Summary: A survey of 1,948 adults in England and Wales asked how likely they would be to report crimes of escalating severity. Using clustering, the study identified three distinct 'reporting profiles'. Willingness to report varied with income, education and employment, and lower-harm offences were interpreted far more unevenly than serious ones - particularly among more disadvantaged groups - suggesting police-recorded crime is a socially filtered, partial map of actual harm. You'll be interested if: you work with crime data, demand analysis or community confidence, and want a reminder that recorded crime reflects who chooses to report as much as what happens, with equity implications for how lower-harm offences are counted. Measuring macro and micro geospatial changes in New York City's motor vehicle thefts Authors: Cubitt, Connealy & Sherman Journal of Experimental Criminology Study design: Geospatial analysis using a natural experiment (before and after a policy change) Setting: United States Summary: This study introduces a spatial analysis method - 'Distance of Distances' - for detecting clusters of crime, and tests whether it can track changes in where crime happens after a major disruption. Using New York City's COVID-19 'shelter in place' period, the authors examined shifts in motor vehicle theft both citywide and around acute-care hospitals, demonstrating the method's ability to capture both macro and micro geographic change. You'll be interested if: you work in crime analysis, hot-spot targeting or performance measurement and want a new technique for spotting how the geography of crime shifts after a policy change or major event. Domestic abuse victimisation in a police workforce: victim experience and the institutional response Authors: O'Leary, Brennan & Couto CrimRxiv (preprint — not yet peer reviewed) Study design: Qualitative study (analysis of open-ended survey responses) Setting: England Summary: This study examines the experiences of domestic abuse victim-survivors who work in policing - an under-researched group facing a clear role conflict between being victims and being part of the institution that responds to abuse. Through qualitative analysis of open-ended survey responses from staff in an English force, it explores how victims experienced the abuse and their force's institutional response, and the private and professional harms that can follow. Note: this is a preprint and has not yet been peer reviewed. You'll be interested if: you have responsibility for workforce wellbeing, professional standards, HR or the VAWG agenda, and want to understand how officers and staff who are themselves victims of domestic abuse experience reporting and support within their own organisation.

Knife crime remains one of the most visible and emotive challenges facing policing in England and Wales. But behind the headlines, what does the evidence actually tell us about who is at risk of escalating from carrying a knife to using one - and what does that mean for how policing and partners respond? That was the focus of the tenth episode of SEBP's Easier Said Than Done webinar series, delivered in partnership with the Youth Endowment Fund on 16 June 2026. The panel brought together Prof. Simon Harding, Director of the National Centre for Gang Research and Criminology Services, and Detective Inspector Paul Connelly of the NPCC National Knife Crime Centre, with the discussion chaired by SEBP Chief Operating Officer Matthew Bland. Understanding the landscape Simon opened by situating knife crime across three broad settings: the domestic environment, the night-time economy, and public space, with social media increasingly operating as a fourth. Alongside this, he noted a significant and underreported development in the data: knife crime offences in England and Wales fell from around 54,500 in 2024 to just over 49,000 in 2025, returning to pre-COVID levels. Several major forces recorded year-on-year falls of around 15 to 17%. Simon offered a hypothesis for this decline that practitioners may not have considered: the post-COVID closure of some 500 nightclubs and 2,000 pubs, combined with the shift to app-based transport and food delivery, has reduced the situations in which spontaneous violence occurs. Fewer people queuing in busy taxi ranks or takeaways means fewer opportunities for confrontations to escalate. Whether this explanation holds up to rigorous analysis remains to be tested. DI Connelly added an important clarification on what knife crime data actually captures. Many members of the public assume the headline figure reflects knives taken off people - it does not. Possession of a knife does not feature in ONS knife crime statistics. The recorded figures include threatened use, perceived threat, and incidents involving sharp instruments other than knives. Understanding what the data does and does not capture matters for how forces analyse and respond to the problem. What drives escalation in the street environment The panel's sharpest discussion focused on the urban street context - the environment most closely associated in public perception with knife crime among young people. Simon, drawing on over 500 interviews conducted through his expert witness work, described a highly status-conscious world in which affiliation to a street gang or drug network creates what he called a cast-iron imperative to carry. For those day-to-day members of such networks, not carrying is itself a risk. The triggers for escalation are relational and reputational - perceived disrespect, challenges to hierarchical position, or threats to status, monitored constantly in person and online. In some networks in major cities, Simon described what he terms a stab-on-sight imperative: spontaneous attacks on rivals driven by perceived grievance rather than planned operation. Paul reinforced the point about the highly localised and contextual nature of the problem. There is no one-size-fits-all picture of knife crime. The problem in a South London estate is structurally different from the challenge in West Yorkshire. Before reaching for a solution, the discipline of accurately understanding the local problem is essential, and that analytical capability, both panellists agreed, is inconsistently available across forces. The practice gap One of the most pointed moments in the session came when Simon offered a direct critique of the current operational picture. Intelligence and data analysis capacity has declined significantly compared to even a decade ago, he argued. Officers are often paralysed by GDPR anxiety and reluctant to share data, despite the legal frameworks for doing so having been in place since the Crime and Disorder Act. Analysts, Simon suggested, have become so rare they risk being studied by David Attenborough. Paul did not dispute the pressure. Policing, he acknowledged, is largely reactive at a local level. The resource required for the kind of sustained preventative analytical work both panellists described simply is not consistently in place. Violence Reduction Units represent an attempt to build that capacity, but Harding was candid that the evidence for their impact remains limited, with too many short-term initiatives, insufficient evaluation, and a tendency toward what he called dynamic inertia - movement without measurable progress. Multi-agency working and where the evidence points Both panellists were aligned on the central conclusion: knife crime cannot be policed away. Paul was direct that policing typically arrives at the end of a young person's story, long after the conditions for violence have taken hold. Education, healthcare, social services, community organisations and the third sector all have roles that policing cannot and arguably should not seek to fill. The Clear-Hold-Build framework was discussed as a structured approach to place-based intervention, combining an enforcement phase with community stabilisation and longer-term investment. Both panellists acknowledged the build phase as the hardest to sustain, dependent on community trust, continued investment and partner agencies maintaining presence after police step back. On focused deterrence, Simon offered a measured assessment. The model has shown promise in specific US contexts but has not transferred cleanly to England and Wales, partly because the compulsory diversion element that underpins the approach in America cannot be replicated here without court orders that our system does not routinely impose. What practitioners can take away Simon's closing observation has stayed with us: we need to find solutions with young people, not solutions to young people. Alongside that, Paul's challenge to parents was characteristically direct - if every parent checked under their child's bedroom tonight, knife crime would fall tomorrow. Neither observation is sufficient on its own. But together they point to something the evidence consistently supports: reducing knife violence requires honest analysis of the local problem, sustained multi-agency commitment, and the involvement of communities and young people themselves in designing what comes next. The recording of Easier Said Than Done 10 is now available. [Watch here — https://www.sebp.police.uk/eventlibrary] The Society of Evidence Based Policing is a UK charity that promotes the use, production and communication of evidence in policing. Use | Produce | Communicate | Evidence

By Dr. Matt Bland TL;DR: Policing is making real progress on responsible AI. Legality, fairness, transparency, governance and public confidence are now being taken seriously. But those safeguards do not quite answer a different question: whether AI tools improve the policing problems they are brought in to solve. The next step is a shared evidence system: common reporting standards, a useful register, synthesis of findings and ongoing monitoring. That would let forces learn not only whether AI tools work technically, but whether they help in practice. Artificial intelligence is arriving in policing very quickly. Across England and Wales, forces are trialling everything from policy assistants and automated redaction to intelligence analysis, risk assessment and live facial recognition. The attention it draws, from chief officers, government, researchers and the public, is intense. What strikes me is that, this time, policing is not walking in blindly. The College of Policing's work on responsible AI, the creation of Police AI, the National Police Chiefs' Council's Covenant for Using Artificial Intelligence in Policing and a long list of exploratory projects all point to a profession trying to think before it deploys. Legality, transparency, accountability, fairness, governance and public confidence are getting serious attention. Recent decisions to pause some uses until they can be properly assessed also show a service that is not willing to be seduced by automation without due diligence. That is real progress, no two ways about it. Historically, policing innovations have often travelled faster than the evidence needed to understand them. The conversation around AI is already more mature than that. But there is a question sitting underneath the responsible-AI agenda that I do not think we are asking clearly enough. Not whether AI should be governed, nor whether it should be transparent or fair. Whether, once a tool is in use, policing will actually learn what effects it has, and whether that learning will be useful to the next force facing the same decision. Working is not the same as helping The distinction matters, because almost everything else in this piece revolves around it. Most of the scrutiny AI gets in policing is about performance: is the system accurate, is it lawful, is it biased, does it produce false matches, and can its outputs be trusted? That work is essential, and a lot of it is being done. Live facial recognition instances for example, has been through technical testing, equitability studies, legal challenge and independent academic review. But performance is not the same as effect. A tool can be accurate, lawful and unbiased and still make no difference to the problem it was bought in to solve. It can save time while weakening the quality of decisions. It can work beautifully in a lab test and struggle in a busy custody suite. It can help one group of people while creating new burdens for others. Knowing that a system works is not the same as knowing that it helps. That second question, whether the tool improves the thing it was meant to improve, is not really a technical or governance question. It is an evidence-based policing question. And it is the one we are not yet answering in a way that accumulates systematically. So, to be clear about scope, this is not a piece about false positives, accuracy or bias. Those issues are well covered elsewhere. This is about effects: whether AI tools are solving the problems they were introduced to solve, for whom, under what conditions, and at what cost. It is worth being precise here, because policing has done good work that is easy to mistake for the answer. The PROBabLE Futures and NPCC Responsible AI Checklist, now embedded in procurement guidance and increasingly a condition of funding, asks whether a tool is technically valid, whether it supports accurate and relevant decisions, and whether its use is lawful and proportionate. Its own framing is telling: it is about deciding whether a tool should be used, rather than whether it could be. Those are exactly the right questions to ask before something is switched on. But they do not answer the question that can only be answered afterwards: did it actually work on the problem? The checklist gates responsible use. It does not (and was never meant to) evaluate effect. The problem is not too little evaluation. It is fragmentation. At first glance this might sound like a plea for more evaluation. It is not. Plenty of evaluation is happening. The trouble is that it appears through a sprawl of pilots, operational trials, academic papers, internal reports, vendor assessments and consultancy reviews: different methods, different outcomes, the same outcomes measured differently, and findings written up in formats that do not easily transfer across audiences. That diversity is not a weakness in itself. Different questions need different methods. An ethnographic study of how officers actually use an AI assistant can reveal things a controlled trial never will; an operational pilot can surface risks technical testing misses. The problem starts when the method, the context and the limitations are not described clearly enough for anyone else to work out what was tested, judge whether the findings might travel, or test them somewhere new. Consistent reporting will not make findings universally generalisable. Nothing will. But it would let a leadership team approaching a vendor with a pressing problem make use of what others have already learned, rather than starting from scratch or taking the sales deck on trust. As the volume of AI activity grows, policing faces a very old challenge in a new form: making knowledge accumulate rather than fragment. What medicine worked out Other fields have been here. Medicine did not fix inconsistent evidence by forcing every study into the same mould. It developed shared standards for reporting different kinds of research. The CONSORT Statement sets minimum reporting standards for randomised trials; STROBE does the same for observational studies; and AI-specific extensions such as CONSORT-AI spell out the extra detail needed to make sense of studies involving AI: how the system was used, how inputs and outputs were handled, how humans interacted with it, and how errors were analysed. These standards do not dictate the research question or guarantee that a study is good. They do something simpler and more useful: they make it possible to understand what was done, decide how much weight to put on it, and set it alongside other studies. I would strongly suggest it is not for policing is not to copy clinical trials. Most police AI evaluations will look nothing like them. The lesson is perhaps that evidence becomes useful when it is reported consistently enough to be scrutinised, reused where it fits, and combined with other evidence. Reporting is not the end of the process - it is the foundation everything else stands on: Evaluation generates the evidence. Reporting makes it interpretable. A register makes it discoverable. Synthesis turns it into shared knowledge. And decision-makers use that knowledge, alongside legal, ethical, financial and operational judgement, to decide whether to adopt, scale, change or stop. Take any one of those links out and the chain doesn’t hold. A test case: live facial recognition Live facial recognition is the fairest demonstration of this argument I can think of, because it is the most heavily scrutinised AI-enabled intervention in English and Welsh policing. If evidence should accumulate anywhere, it should accumulate here. A quick search turns up several pieces: the 2018 evaluation of South Wales Police's use of automated facial recognition, produced through the Universities' Police Science Institute at Cardiff; the 2019 independent review of Metropolitan Police trials by Fussey and Murray, held in the University of Essex repository; the 2023 National Physical Laboratory equitability study on technical performance and demographic differences; the 2020 Court of Appeal judgment in Bridges on legality; force websites with deployment records; regulators and civil-society bodies with further analysis. That is not an absence of evidence. It is an evidence trail, scattered across university repositories, force websites, technical reports, case law and campaign groups. And notice the range of coverage. One covers accuracy. One covers demographic difference. One covers legality. One covers operational deployment. Almost all of it is about performance or compliance. None of it, alone, answers the question a chief officer actually has to decide: is live facial recognition a worthwhile investment compared with the alternatives? For a technology this expensive, intrusive and resource-hungry, that means coverage, opportunity cost and additionality: what share of relevant places and times can realistically be reached, how many people are identified who would otherwise have been missed, what happens after a match, and how the spend compares with putting the same money into intelligence, neighbourhood policing or investigation. The evidence to start answering those questions exists in pieces. What is missing is a front door: a single catalogue entry describing the question, the system and version, the setting, the method, the outcomes and the limitations. When I searched the College of Policing's What Works Crime Reduction Toolkit (in June 2026), live facial recognition was not there as a catalogued intervention. The Toolkit has a particular remit and this is not a criticism of it, but it does show the gap. A register could be that front door. Not a sprawling national database, just checklist-compliant metadata, links to the source material and a clear note of any access restrictions, open where it can be and controlled where it must be. Encouragingly, a national, public-facing registry is now being built as part of policing's new AI centre, which makes this less a call to invent something than a chance to shape what it captures. Catalogue evidence consistently and lightweight synthesis tools, built fittingly with AI, could find comparable evaluations, summarise their findings and limitations, and show honestly where the evidence is too thin to support a confident answer. Not to replace expert judgement, but to make a fragmented evidence base navigable. The facial recognition case makes one last point. The thing being evaluated is never just the algorithm. It is the whole intervention in use: the technology, the threshold, the users, their training, the workflow, the local policy and the human oversight wrapped around it. Catalogue that, or you never really know what you are looking at. A register is coming, but a register of what? This is the part I most want to get right, because policing is already moving and the worst thing this argument could do is ignore that. A national, public-facing registry is on its way. That is genuinely good news. But look closely at what it is for. A transparency registry tells the public that a force is using a tool and how it is governed. The benefits work running alongside it, including economists, cashable savings and hours of officer time released, tells us what a tool costs and saves. Both are important but neither addresses whether the tool change the outcome it was meant to change (unless it was to save money, perhaps). What about questions on crime prevention or investigative outcomes or harm reduction? As far as I can see, nobody is yet asking that systematically across AI-enabled crime interventions. That is the gap, and the register that is coming is exactly the thing to build it into. It also reframes the ownership question. The governance question is largely settled: there is a body that owns the responsible-AI requirement across policing, and a funding and procurement gate that increasingly enforces it. Its current centre of gravity, reasonably enough, is productivity and back-office AI, where credibility is quickest to demonstrate. So the open question is narrower and more important than 'who governs AI?' It is: who owns the synthesis and stewardship of effect-evidence across the whole span, from productivity tools to outward-facing crime interventions, where effect and public legitimacy are most tightly bound, so that what one force learns is usable by the next? Not more money, necessarily, but an expectation A coherent evidence base will not appear through individual research incentives and goodwill. It needs an owner who keeps the standards current, assures the quality of what goes into the register, and turns accumulated findings into a plain 'so what?' for busy practitioners. And it needs consequences. The tempting lever is money: make compliant reporting a condition of central funding. I am wary of that. Most police AI initiatives are not centrally funded. They are bought locally, so the funding lever reaches only part of the problem. But the machinery for the better lever already exists. Responsible use is already becoming the gate for funding, procurement and approval through the Responsible AI Checklist. The same gate could carry a clear expectation that the effects of AI interventions are evaluated and reported to a common standard, distinct from and additional to the performance and bias testing the checklist already covers, until it becomes muscle memory rather than a new bureaucracy bolted on the side. That expectation has to be proportionate. An AI assistant for navigating HR policy should not carry the same evidential weight as a risk-assessment tool shaping safeguarding decisions, or a facial recognition system that can lead to someone being stopped in the street. The answer is a shared minimum: a small set of questions every evaluation should answer, with heavier requirements layered on for higher-risk uses. At a minimum, I’d want every evaluation to be clear about: what problem the tool was meant to solve what system and version were tested how it changed the existing process who used it, where, and under what conditions what baseline or comparison was used which outcomes, costs and potential harms were measured how human judgement interacted with the system whether effects differed between groups or contexts what failed, and what surprised everyone whether the system or the environment changed mid-evaluation Not methodological uniformity. Just enough consistency to learn. The opportunity, and the catch AI does give evidence-based policing a real opportunity. These systems generate a detailed data footprint as they run: usage, outputs, corrections, response times, error rates and what happened next. Access can sometimes be phased between sites, or interfaces compared through A/B testing where that is lawful and appropriate. Some AI interventions are, as a result, unusually measurable, and policing could learn from them much faster than from innovations whose effects are almost invisible. But measurable is not the same as easy. The data may be incomplete, owned by the vendor, or only loosely connected to anything that matters. An A/B test can tell you which version gets used more without telling you which produces better policing. Models update, prompts change, performance drifts, and the most important harms are often rare, delayed or hard to count. That is exactly why effect-evaluation has to be planned from the start, not bolted on at the end, and why it needs literate customers. If leaders, practitioners and commissioners cannot judge what a system does, where it fails and which vendor claims deserve testing, policing lets the people selling the tools define both the problem and the proof of success. Why bother, isn't this just more bureaucracy? Very few members of the public spend their evenings thinking about intelligence triage, records management or disclosure tools. Yet those unglamorous systems shape the quality of policing profoundly, and we usually only notice when something goes wrong. Nobody campaigns for better intelligence triage until a dangerous offender is missed; nobody worries about a risk-assessment tool until a safeguarding failure; nobody thinks about disclosure support until a miscarriage of justice. If investment follows visibility, novelty and political heat, policing will keep underinvesting in the quiet tools that do real good and overinvesting in the ones that catch the eye. The antidote is to make far better use of what policing already knows. There are deep reserves of insight in forces up and down the country, and too much of it is stuck in procurement files, internal reports and local pilots. Much of it is effectively lost. The result is a service where outcomes are shaped, in part, by postcode. For a profession that has spent three decades arguing that evidence should inform policing, that should sit uncomfortably. So, to be clear about what I am not arguing for: not another governance checklist to be completed and filed before go-live, and not a demand that every deployment clears a randomised controlled trial first. What policing needs is a shared evidence architecture that supports learning across the whole life of a system. Five parts: A common minimum reporting standard for the effects of AI interventions, informed by CONSORT, STROBE and CONSORT-AI but built for policing. It should be demonstrable through a short checklist, without pretending that the checklist turns the underlying work into peer-reviewed research. Risk-tiered, use-case-specific guidance on method, so higher-risk uses face more demanding requirements and low-risk admin tools are assessed proportionately. Effect built into the register that is coming. The national registry now in development is the right foundation, but a registry that records that a tool is used and how it is governed is not the same as one that records whether it worked. It needs to carry evaluation findings, including the negative and inconclusive ones, open where possible and controlled where necessary. The government's Algorithmic Transparency Recording Standard is a useful precedent, but transparency about a system is not evidence about its effects. A synthesis and stewardship function that keeps the standards current, assures the register’s quality, produces usable summaries and says plainly what the evidence does and doesn’t support. Continuing monitoring after deployment, because models, behaviour and environments change, and a pilot’s reassurance has a shelf life. None of this is free of hard questions: commercial confidentiality, operational sensitivity, data protection, the cost of evaluation, and the limits of generalising from one force to another. But those are reasons to design carefully, not reasons to settle for fragmented learning. Policing has set itself the goal of becoming a more intelligent customer of AI. I would gently point out that you cannot be an intelligent customer of something whose effects you never measure. Evidence-based policing has spent more than thirty years making the case that scientific method can improve policing. AI is about to increase both the pace of innovation and the chance to evaluate it. This is the moment to build the habits and the plumbing that let the public benefit from what policing learns, not just about whether its tools work, but about whether they help. References and further reading Algorithmic Transparency Recording Standard: guidance for public sector bodies CONSORT Statement CONSORT-AI: reporting guidelines for clinical trials involving artificial intelligence DECIDE-AI: reporting guideline for early-stage clinical evaluation of AI decision-support systems NIST AI Risk Management Framework STROBE Statement PROBabLE Futures and NPCC, Responsible AI Checklist for Policing (Oswald, Calder, Paterson-Young and Dunkwu), May 2025 NPCC, Artificial Intelligence Playbook for Policing, 2025 National AI centre for policing (Police.AI), launch announcement, including a public-facing AI registry Davies, B., Innes, M. and Dawson, A. (2018), An Evaluation of South Wales Police's Use of Automated Facial Recognition, Universities' Police Science Institute / Cardiff University. Fussey, P. and Murray, D. (2019), Independent Report on the London Metropolitan Police Service's Trial of Live Facial Recognition Technology, University of Essex Research Repository. National Physical Laboratory (2023), Facial Recognition Technology in Law Enforcement: Equitability Study, commissioned by the Metropolitan Police and South Wales Police. R (Bridges) v Chief Constable of South Wales Police [2020] EWCA Civ 1058, Court of Appeal judgment. College of Policing, Crime Reduction Toolkit, searched for live facial recognition as a catalogued intervention, June 2026. Metropolitan Police and South Wales Police live facial recognition deployment records and published operational results.

Thursday 12th November 2026 | Online via Microsoft Teams | 09:30–16:30 The Society of Evidence Based Policing (SEBP) Midlands Region is inviting submissions for presentations, research spotlights, and practice-sharing sessions at our 2026 annual conference. This year's online conference will explore two key themes: Morning session – AI in Policing We welcome submissions on the use of artificial intelligence across policing, from predictive tools and decision-support systems to the ethical and practical challenges of implementation. We are particularly keen to hear from both researchers and practitioners with frontline or operational insights. Afternoon session – Problem-Solving We are looking for submissions that showcase evidence-based approaches to problem-solving in policing contexts, including case studies, evaluations of local initiatives, and research that bridges theory and practice. Who should submit? We welcome proposals from police officers and staff at all ranks and grades, academics, researchers, and partner agencies across the Midlands region and beyond. Whether you have a completed study, an ongoing project, or a practice-based case study to share, we want to hear from you. What to include in your submission: Your name, rank/role, and organisation Title of your presentation Which session theme it relates to (AI in Policing / Problem-Solving) A short abstract (max 250 words) summarising your topic and key findings or discussion points Preferred format (presentation, panel discussion, Q&A) Submission deadline: 7th August 2026 Submit via the Microsoft Form: Call for Submissions - SEBP Midlands Region Conference 2026 – Fill in form Places at the conference are free and open to all. Register now at: https://www.sebp.police.uk/events-1