March 2, 2017

The concept and measurement of violence against women and men

This new book proposes a new and coherent measurement framework for violence. It offers a simple set of indicators to assist public policy. Based on decades of research, this group of twelve authors, drawn from a wide range of administrative and survey backgrounds, have reached agreement on a new framework.

Existing data is fragmented and uneven, using different definitions and units for measurement. Surveys and administrative data are so different that they cannot be meaningfully compared.  Categories for data on violence in the criminal justice system are different from those in health and are different from those used by specialised services and all are different from those in surveys.  Even the two parts of the Crime Survey for England and Wales use different categories.

Including gender in official statistics is just beginning. Still, most recorded crime statistics do not include the gender of the victim, thus making them unusable to answer the question of how much violent crime there is.  Even the new International Classification of Crime for Statistical purposes from the United Nations Office of Drugs and Crime, has this problem.  Some new approaches focus on women only, making comparisons with men impossible.  Gender is more than the sex of the victim: it involves the nature of the relationship between perpetrator and victim.  This book offers a new approach: mainstreaming a multi-dimensional understanding of gender into statistics on violence.

Repetition of violence is gendered. The assumption of ‘one victim, one perpetrator, one offence’, is gender-biased.  This is because the forms of violence that are repeated are the ones that are most often perpetrated against women: domestic violence in particular.  Even though high rates of repetition of violence against women are collected in surveys, such as the Crime Survey for England and Wales, traditionally these are not all included in the estimates.   This leads to an underestimate of the extent to which violent crime is committed against women (as compared with men) and of the extent to which violent crime is perpetrated by people known to the victim (domestic relations or acquaintances) rather than by strangers.  Repetition needs to better addressed in the measurement framework than has been conventionally achieved.

The measurement framework requires consistent and coherent conceptualisation of ‘violence’ and of ‘gender’.

The new measurement framework


  • Actions (and intentions) and harms (and non-consent)
  • Variations by types of violence

 Gender dimensions

There are five gender dimensions:

  • The sex of the victim
  • The sex of the perpetrator
  • The relationship between perpetrator and victim: whether the perpetrator was a domestic relation – either (current or former) intimate partner of the victim or another family member (either blood relative or other household member), an acquaintance or a stranger
  • Whether there was a sexual aspect to the violence as well as physical (potential, not necessary dimension)

Unit of measurement

Three units of measurement need to be used at the same time:

  • Event (incident, crime, episode and so on)
  • Victim
  • Perpetrator

For consistency, further issues need to be addressed:

  • Ages of perpetrator and victim (adult/minor)
  • Temporality (event within the last 12 months; enduring rather than an event)
  • Harmonised standards for moment of definition (reporting, investigation, court decision)
  • Harmonised counting rules (what takes precedence when there are multiple crimes, victims or perpetrators in the same event)


An indicator is a summary statistic that is meaningful to non-experts.

In the field of gender-based violence, administrative data is not a good basis for an indicator for any form of violence other than homicide. This is because an improvement in services (say by the police) will increase the recorded rate of violence at the same time as it drives down the real rate of violence. Only homicide is robust against this perverse effect.

Three indicators are possible: femicide (gender disaggregated homicide), domestic violent crime and conviction rates for homicide and rape. Though even here there is a need to improve data collection.

Femicide: gender disaggregated homicide

Homicide, disaggregated by the sex of the victim and also by whether the relationship between perpetrator and victim was domestic, is potentially available using administrative data. Counting rules could be provided by the United Nations Office on Drugs and Crime (UNODC) International Classification of Crime for Statistical Purposes (ICCS), with some minor revisions, including elevating the sex of the victim, the sex of the perpetrator and the relationship between them to mandatory codes rather than optional tags.

 Domestic violent crime

Violent crime, disaggregated by the sex of the victim and by whether the relationship between perpetrator and victim was domestic (including whether intimate partner or other family member), is potentially available using survey data. Domestic violence is limited to those types of violence that pass the crime threshold, for practical reasons. Relevant data can be gathered by surveys, such as the Crime Survey for England and Wales main questionnaire, (though the CSEW specialised self-completion module lacks the necessary data on frequency).

 Conviction rates for femicide and rape

Conviction rates for femicide and rape are calculable from available administrative data. These would be indicators of policy performance, not of the ‘real’ rate of violence.

Theoretically relevant and pragmatic

These proposals are both theoretically informed and pragmatic. Much of the data required for the proposed measurement framework is already collected, even if it is not adequately abstracted into the statistical summaries.  For example, there is more data in police files on interviews than makes its way into national level statistics.

The benefits of a coherent measurement framework on violence, that makes visible its gender dimensions, would be enormous. It would facilitate the cooperation needed for multi-agency working.  It would aid the evaluation of what works for the commissioning of services in a multi-agency environment.  It would aid the development of the theory of change, which is needed to underpin all efforts to end and reduce violence.

Walby, Sylvia, Jude Towers, Susie Balderston, Consuelo Corradi, Brian Francis, Markku Heiskanen, Karin Helweg-Larsen, Lut Mergaert, Philippa Olive, Emma Palmer, Heidi Stöckl, and Sofia Strid (2017) The Concept and Measurement of Violence against Women and Men. Bristol: Policy Press.

March 2, 2017