Predicting the criminal records of male-on-female UK homicide offenders from crime scene behaviours

Research Institution / Organisation

University of Liverpool

In Collaboration With

National Crime Agency (NCA); Serious Crime Analysis Section (SCAS)

Principal Researcher

Dr Louise Almond

Level of Research


Project Start Date

March 2018

Research Context


  • To investigate possible links between male-on-female homicide offenders’ criminal history and their crime scene behaviours.


  • To apply the methodological approach of Davies, Wittebrood and Jackson (1997) to the context of male-on-female homicide offenders using a contemporary sample from the UK’s National Crime Agency’s ViCLAS database
  • To examine bivariate relationships between crime scene behaviours and previous convictions of male-on-female homicide offenders
  • To identify statistical models that are qualified to predict an offender’s criminal history from combinations of specific crime scene behaviours
  • To interpret and present findings for use in contemporary investigative practice.

Research Methodology

The study aims to draw from secondary data of cases of male-on-female UK homicide offences recorded in the National Crime Agency’s (NCA) ViCLAS database (sexual or unknown motive).

Selection criteria
Crime scene behaviours and previous convictions of male homicide offenders with female victims will be requested from the ViCLAS database according to the following criteria:

  • Offences must have occurred between 1985 and 2017
  • All cases must be solved male-on-female homicides with both offender and victim being adults (above the age of 16)
  • Data must be taken from cases involving single offenders only
  • For serial homicide offenders, only the first victim will be included in the dataset.

Offender, victim and offence behaviours will be coded.

Data will be analysed in two stages. In stage one, chi-square analyses will be used to investigate, whether there are any direct associations between offence behaviour variables and conviction variables. In stage two, logistic regression models will be used to predict an offender’s previous convictions based on a combination of significant offence behaviour variables identified in stage one.

Interim reports and publications

​Not available

Date due for completion

March 2019
Return to Research Map