A problem-solving and analytical approach to serious violence.
Aims
This guide is about problem solving and the analysis of serious violence. It aims to:
- describe a range of practical tools and techniques that you can use to help better understand and respond to problems of violence
- cover concepts and theories which have proven useful when applying a problem solving approach to violence reduction
This guide is written primarily for analysts but should be relevant to anyone with an interest in, or responsibilities for reducing violent crime.
How this guide is organised
This guide covers topics relevant to problem solving violent crime. It is divided into five sections corresponding to the SARA problem solving model. These sections are:
- data for problem solving analysis
- analysis to identify and prioritise problems
- analysis to determine patterns
- analysis to evaluate impact
- presenting analysis effectively
Each section is made up of units. Each unit covers three areas:
- the learning objectives of the unit
- a description of why the unit is important when problem solving
- a demonstration, drawing on research and practice, of how the information reported in that unit might assist you in problem solving violence
Each unit ends with signposts to recommended resources and readings.
Explainer videos
The following explainer videos highlight four of the units about hot spots policing. Each unit is explained further in the download of this guide.
Selecting suitable problems to solve
A problem suitable for problem solving has four general characteristics:
- it must be recurring – not a one-off incident
- those recurring incidents need to be connected in some meaningful way, be that particular places, offenders or victims
- it must negatively affect the community
- it must fall within the police remit to do something about
Many of the issues that the police are routinely called upon to handle meet these four criteria, including violence against women and girls, anti-social behaviour and serious violence. However, there is often value in digging deeper into these broad categories of offences.
There is value in doing so because these broad categories of crimes often cover several different and distinct problems, which may involve different people, with different motivations giving rise to different crime patterns.
How you deal with those problems may require different responses working alongside different partners.
Problem solving emerged as a reaction to the predominantly reactive model of policing that dominated at the time. It called on the police to focus not only on responding to individual incidents, but also to the prevention of repeat sources of demand, so-called persistent problems.
But what exactly counts as a ‘problem’ for the purposes of problem solving – and which problems are best suited to this approach?
A problem suitable for problem solving has four general characteristics. First, it must be recurring – not a one-off incident. Second, those recurring incidents need to be connected in some meaningful way, be that particular places, offenders or victims. Third, it must negatively affect the community. And fourth, it must fall within the police remit to do something about.
And for this reason, problem solving is less about tackling broad social forces like poverty and inequality, which contribute to crime, but which sit largely outside of police control.
Now, many of the issues that the police are routinely called upon to handle meet these four criteria. Violence against women and girls, anti-social behaviour, serious violence, they all meet these criteria.
However, for the purposes of problem solving, there is often value in digging deeper into these broad categories of offences. And there is value in doing so because these broad categories of crimes often encompass several different and distinct problems, which may involve different people, with different motivations giving rise to different crime patterns.
Moreover, how you deal with those problems may require different responses working alongside different partners.
Take knife crime. Knife crime is routinely discussed as a single problem – an example of serious violence. But in reality, it encompasses a range of quite different issues, from knife-enabled robbery to knife-enabled assault to the illegal sale of certain knives.
And even within these sub-categories, there may be important differences. Knife-enabled robbery of school children for their mobile phones is a rather different problem to knife-enabled robbery targeting expensive watches. It occurs at different places, at different times, it involves a different type of victim and, crucially, it may require different police and partner responses.
Going too broad when problem solving often leads to a shallow analysis and generic responses. Being specific reveals the patterns that might point towards effective solutions. Generally, the tighter you define your problem, the sharper your response can be.
Hot spots and hot times
Effective crime prevention depends on precision. Not just knowing that crime happens, but exactly where and when it happens. Understanding hot spots and hot times changes how we target resources and develop responses.
A hot spot is an area with high crime concentration relative to the surrounding region. Hot spots exist at different scales – from neighbourhoods down to specific streets. Hot spots, however, are not hot all of the time.
Combining where and when results in better analytical precision and how resources can be targeted. Rather than spreading patrols across entire areas or generalised time periods, you can target precise locations during the specific hours when crime concentrates. This means that limited resources can be deployed where and when they will have the greatest impact.
Effective crime prevention depends on precision. Not just knowing that crime happens, but exactly where and when it happens. Understanding hot spots and hot times transforms how we target resources and develop responses.
A hot spot is an area with high crime concentration relative to the surrounding region. Hot spots exist at different scales – from neighbourhoods down to specific streets.
We know crime concentrates. What's striking is the extent. Research consistently shows that around1% of places in a city account for about 25% of all crime. Roughly 5% of places account for half. This is the law of crime concentration. It means that effectively targeting hot spots can make significant inroads into an entire city's crime problem.
Techniques such as those that identify and map the streets that account for these high percentages of crime – kernel density estimation and the Gi* statistic – can be used for showing where crime concentrates.
Hot spots, however, are not hot all of the time. When crime happens is not random. It also concentrates. Data clocks show this clearly.
Analysis of robberies in Newcastle found that while Fridays had the most incidents overall, they clustered in specific windows: late afternoon and again late evening. Violence in Hull showed that while Sundays had the most incidents, these took place in the early hours and were mainly associated with the Saturday night-time economy.
Combining where and when results in better analytical precision and how resources can be targeted. Rather than spreading patrols across entire areas or generalised time periods, you can target precise locations during the specific hours when crime concentrates. This means that limited resources can be deployed where and when they'll have the greatest impact.
Identifying this precision also prompts the question, why at this place, at this time? That is, what is making this location a hot spot?
Crime has patterns in both space and time. Finding those patterns is where effective problem solving begins.
Risky facilities
Identifying crime patterns is an important part of effective problem solving. When we analyse crime, we tend to focus on where and when crime happens, but these are just two forms of crime patterns. There are others worth examining – such as how crime is distributed across similar types of places. This is known as a ‘risky facilities’ analysis.
Think of a facility simply as a location that serves a particular function, such as:
- bars
- hotels
- hospitals
- libraries
- petrol stations
- airports
In a risky facilities analysis we investigate how a particular crime is distributed across similar sets of facilities in a given area. This could be a particular town or maybe an entire police force area.
Doing so often reveals a consistent pattern – crime is not spread evenly across facilities. Most experience little to no crime, yet a small number of risky facilities experience a lot. Together those risky facilities account for a large share of all crime taking place at those locations.
Crime is highly patterned. Identifying those patterns is a key part of effective problem solving. When we analyse crime, we tend to focus on where and when crime happens. But these are just two forms of crime patterns.
There are others worth examining – such as how crime is distributed across similar types of places. And this is known as a ‘risky facilities’ analysis.
What is a facility? Think of a facility simply as a location that serves a particular function. Bars, hotels, hospitals, libraries, petrol stations, airports – these are all examples of facilities.
Now, in a risky facilities analysis, we investigate how a particular crime is distributed across similar sets of facilities in a given area, be that a particular town or maybe an entire police force area.
Doing so often reveals a consistent pattern: crime isn't spread evenly across facilities. Rather, most experience little to no crime. And yet, a small number of risky facilities experience a lot, and together those risky facilities account for a large share of all crime taking place at those locations.
Consider violence in the night-time economy. In a risky facilities analysis, you would examine how violence is distributed across all licensed premises in a given town or city.
Now, many analysts have done this and the same general pattern emerges. Most pubs and bars experience few incidents of violence, and a small handful experience a lot.
In a recent study from one UK city, across 72 licensed premises, just nine accounted for half of all the violence taking place across those venues.
Now why is a risky facilities analysis important? It's important for two reasons. First, it helps with resource allocation. Rather than spreading prevention efforts thinly across all venues, you can focus your resources on those small number of facilities that cause the most harm.
Second, it prompts useful questions. By identifying the risky facilities where crime concentrates, it begs the question: why does my crime problem concentrate at those locations and not those [other] locations? What is it about these facilities that makes it conducive to crime?
Making these comparisons can often yield important insights which might then point towards effective solutions.
Control groups and the weighted displacement difference (WDD) test
Evaluation rests on a simple idea. If we could create two identical situations – one where we intervene and one where we do not – any difference in the outcome would be down to our intervention.
To do this we create a counterfactual – a comparable area or group that received business as usual while the intervention area receives our programme. If we are confident the control reflects what would have happened without intervention, we can make stronger claims about what our action achieved.
The WDD test is a straightforward technique for measuring whether a geographically targeted intervention produced a statistically significant change in crime. It compares crime counts across two areas:
- the treatment area where you intervened
- a control area for comparison
Control groups tell us what would have happened without action. The WDD test quantifies the difference our action made. Together, they turn hopeful observation into credible evidence.
How do we know if an intervention actually worked? We can see crime fell after we acted – but would it have fallen anyway? Answering this requires comparison groups and robust measurement. Here's how to do both.
Evaluation rests on a simple idea. If we could create two identical situations – one where we intervene and one where we don't – any difference in the outcome would be down to our intervention.
We can't be in two places at once, so we create a counterfactual: a comparable area or group that received business as usual while the intervention area receives our programme. If we're confident the control reflects what would have happened without intervention, we can make stronger claims about what our action achieved.
The challenge lies in ensuring those comparison areas are comparable – similar risk factors, similar conditions, no spillover effects between them. No comparable area is ever perfect, but it's better to have at least one control area to compare to than nothing at all.
The weighted displacement difference test – or WDD – is a straightforward technique for measuring whether a geographically targeted intervention produced a statistically significant change in crime. It compares crime counts across two areas: the treatment area where you intervened and a control area for comparison.
A third area can also be included if data is available – a buffer zone surrounding the treatment area to capture any local displacement. The calculation accounts for changes in all three.
If crime dropped in your treatment area but also dropped equally in the control area, that suggests something else was driving the change. Including the buffer zone captures neighbouring changes in crime to see if displacement has occurred or if the intervention in the treatment area produced a diffusion of benefit effect to the surrounding areas. The WDD isolates the effect attributable to your intervention.
Consider a robbery intervention. Before the programme, the treatment area had 33 robberies. Afterwards, it had 19. That looks promising – but the control and buffer areas also saw changes. Running the WDD calculation produces a result of minus 20, meaning a net reduction of 20 crimes, accounting for changes elsewhere. The statistical test incorporated in the WDD confirms this result is significant. That is, it's unlikely to have occurred by chance. This gives confidence the intervention genuinely reduced robberies, beyond what would have happened anyway.
Control groups tell us what would have happened without action. The WDD test quantifies the difference our action made. Together, they turned hopeful observation into credible evidence.