The effects of Police Community Support Officer hotspot foot patrol on crime and incidents in a small city centre setting

Randomised Control Trial Status


Trial Hypothesis

The key research question this study sets out to answer is;

  1. Do hotspots in receipt of foot patrol by uniformed Police Community Support Officers experience reductions in community-reported crime and incidents when compared to those to which no foot patrols are assigned?

The research also aims to address the following sub-questions;

  1. What barriers need to be overcome to create effective hotspots in a small city centre setting? 
  2. Across a 90-day experiment, what are the differences between patrol minutes for the treatment and control hotspots?
  3. What activity do Police Community Support officers complete whilst on patrol?
  4. What conclusions and recommendations can be made about hotspot patrol designs for smaller city centre settings?

Geographical area

​Sheffield City Centre .

Research Institution / Organisation

University of Cambridge

In Collaboration With

South Yorkshire Police

Participants - inclusion criteria

To achieve the aim of identifying micro-place hotspot patrols the following definition of a hotspot was adopted for this study;

  1. An area no bigger than 50 meters x 50 meters
  2. With no less than 200 community reported crime or incident in the past 24months (excluding Shoplifting and Police initiated events) .

To identify such hotspots, crime and incidents data from the 1st of May 2016 to the 30th of April 2018 was mapped over the Sheffield city centre area. To reflect the timing of the patrols conducted by PCSOs this data was limited to those occurring between 0800 and 2200. This would ensure that the hotspots were relevant to the timing of the patrols and would not factor in offences that occur in the night time economy for example. From this data, shoplifting offences were removed as this skewed the hotspots almost purely to retail premises, whilst police generated crime and incidents (such as drug arrests) were also excluded as these are outputs rather than outcomes (Sherman and Weisburd,1995).

Target Sample Size

​8 high demand micro hotspots were identified for inclusion into the study, which were then matched into 4 pairs based on their levels of crime and incidents. In total these 8 hotspots contained 3924 crimes and 9794 incidents in the 2 years prior to the study.

Study Design

​The research is designed as a randomised control trial using random assignment of treatment and control conditions to pairwise matched hotspots, whilst measuring crime and incident data before and after in multiple experimental and control units.


Those hotspots which were allocated for treatment received 90 days of hotspot foot patrols by PCSOs. These patrols consisted of four 15-minute visits a day between the hours of 0800 and 2200 to match the PCSOs shifts. These timings were based on the work of Koper (1995) who found that the ideal dosage for police patrols at a hotspot was 15 minutes to have the maximum residual deterrence effect over the proceeding 30 minutes. All the PCSOs involved in the experiment were briefed at the beginning of the experiment as to the requirements of their patrols and were given a briefing document when they were assigned to the patrols so that they understood what was required from them. These briefing documents all contained a map of the hotspot for their reference and suggests about taskings or key businesses to visit within the hotspot to try and focus their activity. 

The patrols consisted of two PCSOs patrolling together each shift, 7 days a week. The PCSOs were dedicated to the Hotspot foot patrols throughout the shift and given a patrol route designed around the temporal hotspot analysis that is unique to that weekday. These routes gave a one-hour window in which the PCSOs was expected to make their 15-minute patrol, meaning the actual patrol time will be different each day but focussed on the key times that crimes and incidents occur.

Outcome Measures

This experiment set out to collect and measure the outcome of the patrols based on three key measures; recorded crimes, police incidents and harm scoring. Crimes and Incidents were recorded both in total and per day, as well as being broken down into their individual hotspots to allow for additional analysis regarding each matched pair. Police generated crimes and incidents were again excluded from the outcome measure, as they are described by Sherman and Weisburd (1995) as outputs rather than outcomes and are partially captured in this study through the data analysis of the patrol activity. Harm was measured against the recorded crime using the ONS (2018) severity score so that the experiment could measure any potential change over the experimental period. Harm scores were only calculated for recorded crime, as currently there is no satisfactory way of converting incidents into a harm score. 

The primary outcomes of interest for this experiment was the change between the experimental period and the previous year’s demand comparing the control and treatment hotspots. Data from the experimental period in both 2016 and 2017 was also captured to allow for year on year comparison.

Summary of Findings

The main findings suggest that on average treatment patrol hotspots saw 83.7% (d=0.78) higher levels of crime and incidents per day compared to the control hotspots over the experimental period. When comparing this against the previous two years data, the treatment hotspots had an average increase of 21.9% in crime and incidents whilst control hotspots saw a reduction of -18.29% per day. 

As intended, based on the GPS and self-reported data, treatment hotspots over the experimental period received on average an additional 48 minutes of PCSO patrol compared to the control hotspots, with the average visit being 16 minutes. The hotspots patrols had a compliance rate of 73.2%, and over the study period there were 133 crimes and 246 incidents reported to the police within the hotspots. In 94% of those patrols, activity was focussed on Community Engagement and in only 6% of visits did the PCSO’s use their limited enforcement powers. 

Overall the study showed no evidence that Police Community Support Officer Hotspot Foot Patrols reduce the levels of crime and incidents in hotspot locations compared to random patrol, and suggests they have the opposite effect, and can increase reporting to police. 

The exact reasons for this are outside the scope of this study, however it is suggested it may be due to a lack of deterrent by the presence of PCSOs, who recorded that in 94% of hotspots visits they only engaged with individuals and did not use their albeit limited powers of dispersal or enforcement. Likewise, the result could also be due to an increase in public confidence in reporting to the police due to their regular attendance in the hotspot, especially in a central businesses district.

This study has however demonstrated that Hotspot policing is applicable to small city centre settings. Micro-location hotspots were identifiable with sufficient levels of crime and incidents even across a single square mile and were still give a clear image of which locations which cause the police a disproportionate amount of demand, in some cases up to 12% of all crime and incidents for that specific area.

Date last updated

Thursday 15 August 2019
Return to Research Map