Dear Sylvia and Brian,

Thank you for taking the time to discuss your concerns about the quality and value of the Office for National Statistics (ONS)’s combined measure of domestic abuse, sexual assault and stalking. ONS developed this measure to support the UK Government in tracking progress against its ambition to halve violence against women and girls (VAWG) within a decade. It is the headline metric in the recently published VAWG strategy.

You shared with us an academic paper outlining ten concerns that you identified with the measure:

• It includes crimes against men and boys
• It does not include violence against girls under 16
• It does not include violence by acquaintances and strangers
• It does not include some forms of VAWG such as female genital mutilation, forced marriage and modern slavery
• It is incompatible with the main Crime Survey for England and Wales (CSEW) series because it does not use crime categories
• It is incompatible with the main CSEW series because it does not measure the number of incidents, only the number of victims
• It does not include a measure of seriousness
• It is dominated by domestic abuse less serious than the criminal threshold
• It obscures the size of the gender gap in more serious forms of violence
• It misses some of the most marginalised and vulnerable individuals

We responded to you privately, setting out our views on each issue. This letter summarises our position and clarifies the basis for our judgements.

The data source for the combined measure

Overall, we consider that ONS has used the most reliable data source to produce the combined measure: the self-completion module of the Crime Survey for England and Wales (CSEW). This module is ONS’s preferred source for measuring victimisation for these crime types.

The combined measure is a prevalence measure, estimating the proportion of people who have experienced an incident of one or more of the three crime types. You have argued that an incidents-based measure, derived from information from face-to-face interviews, would provide a better indicator of the scale of these crimes because it captures repeat victimisation and is consistent with the CSEW headline crime estimates.

We agree that both prevalence and incidents-based measures are important for understanding the extent of VAWG, and we recognise the limitations of the self-completion module. However, we also understand ONS’s rationale for prioritising this data source. Domestic abuse and sexual assault are frequently underreported in face-to-face interviews as respondents may be unwilling to disclose sensitive experiences in a home setting. In addition, face-to-face interviews only capture incidents of domestic violence – one component of domestic abuse – and do not cover stalking. As a result, incidents-based estimates may give a misleading picture of the scale of victimisation. These limitations are explained clearly in ONS’s supporting information for the crime statistics.

You also raised concerns about the public availability of incidents data. ONS publishes data on incidents of domestic violence in its Nature of crime: violence data tables, and detailed incidents data are available to researchers through ONS’s Secure Research Service and the UK Data Service. ONS does not publish data on incidents of sexual assault.

The definition of VAWG and the VAWG strategy

Two of the issues in your paper relate to the government’s definition of VAWG and the broader framework of metrics in the VAWG strategy used to track progress. You expressed particular concern about the inclusion of violence against men and boys in the combined measure.

While we can comment on the quality and presentation of individual metrics, it is not our role to determine how policy concepts such as VAWG should be defined. As such, we do not intervene in concerns about definitions in government strategies or policies. Responsibility for defining and measuring VAWG sits with the Home Office.

In developing the combined measure, ONS followed the Home Office definition. The headline metric covers violence against all people, but ONS also publishes separate estimates for women and girls, and for men and boys, allowing users to understand patterns of victimisation by sex.

We would encourage you to raise any concerns about the government definition of VAWG directly with the Home Office. We have shared the contact details for the relevant team.

The Home Office is also responsible for justifying which crime types are included in the combined measure. In our view, the strategy does not provide sufficient detail about the combined measure or the supporting headline and sub-metrics. In particular, it does not clearly explain why certain forms of violence, such as sexual harassment, are excluded from the combined measure, nor does it describe the data source for each metric.

We have written publicly to Home Office setting our expectations regarding transparent communication of the metrics in the strategy. While we welcome the range of metrics included, we have asked the Home Office to consider publishing additional supporting information on the metrics, data sources and data quality to support public understanding.

Future work on the combined measure

You asked whether we would commit to undertaking a formal review of the combined measure. As the measure is undergoing further development, we do not consider this is the right time for a review. However, given its public profile and societal importance, we may consider reviewing it in future. We note that, because the measure is published as official statistics in development, it does not yet meet the same quality standards as other CSEW estimates, which are published as accredited official statistics. ONS is transparent about the status and quality of the measure.

We hope this response is helpful.

Yours sincerely,

Ed Humpherson
Director General for OSR