Understanding homicide: a framework analysis

Based on evidence gaps highlighted in our 2020 Strategic Needs Assessment, London's Violence Reduction Unit (VRU) commissioned the Behavioural Insights Team (BIT) to work with the Metropolitan Police Service (MPS) to create a framework to better understands homicide, how situational and behavioural factors interact, and how we can work in partnership to identify opportunities to intervene earlier.

For this study, BIT looked at 50 homicide cases to develop, test and validate a coding framework designed to be used by researchers and homicide experts with a wide variety of research questions.

This report selected six factors identified as if particular interest to the VRU; Mental Health, Drugs, Alcohol; Gang involvement; Social media and Patterns of escalation. The analysis in this report is only a small part of the project’s potential. The true impact of this work will come from applying the framework to all homicide cases to create a much larger dataset. 

This research developed a coding framework for turning complex case notes into an analysable, standardised dataset. For each homicide case, the framework takes complex information covering key behavioural and situational factors and streamlines these into common variables.

The variables within the framework were initially identified through a literature review and conversations with homicide experts, to ensure they would contribute to our current understanding of homicide.  

Cases were selected based on prespecified characteristics so the framework could be widely applied. This means that at this stage the sample was not intended to generate a statistically representative sample of homicide cases in London. However, emerging insights provide a flavour of the opportunities going forward, for example: 

• Analysis around mental health issues underlines the varied role it can play in homicides, with individual’s vulnerabilities interacting with each other. For example, worsening mental health may be due to an attempt to stop consuming substances. Analysing these cases in the round means these complexities can be better explored and addressed.  
• In terms of intervention points, the sample analysis suggests potential opportunities for de-escalation – for example, police are often involved in lead up to alcohol-driven homicides.

Crucially, the report demonstrates the validity of the approach, highlighting a wide range of future opportunities as the dataset is expanded and coding is incorporated into routine practice in the recording of homicides in London.

These include:

• potential around machine learning to aid investigation
• data systems to facilitate more efficient decision making
• making the dataset available to a wider research community
• roll out across other police forces
• joining data from a range of other organisations.

The report made several recommendations relating to utilising the framework for homicide cases going forward. The project has now progressed to its second phase, which involves applying the framework to a further three years of cases and embedding the coding process into business as usual. 

The analysis and insight that this opens up will help practitioners, police officers, policymakers, and researchers better understand the drivers of homicide, and ultimately reduce it.