Email correspondence to OSR (16 March 2026)
Dear Office for Statistics Regulation,
Please find attached a formal concern regarding the use of police recorded crime statistics in Violence Against Women and Girls (VAWG) policy documents, including the VAWG Strategic Threat and Risk Assessment (2024) and the National Policing Statement on VAWG (College of Policing/NPCC, July 2024).
The concern relates specifically to the use of police recorded crime figures as measures of crime prevalence, in direct contradiction of ONS guidance on the appropriate use of this data. The submission sets out the relevant ONS and OSR guidance, identifies the specific policy documents and claims in question, and poses a number of questions for the OSR’s consideration.
I am an independent analyst with no institutional affiliation. I am happy to provide any further information or clarification that would assist your consideration of this matter.
Concern Regarding Misuse of Police Recorded Crime Statistics in Violence Against Women and Girls (VAWG) Policy
1. Summary of Concern
This submission raises a concern about the systematic misuse of police recorded crime statistics as measures of crime prevalence in official VAWG policy documents. The Office for National Statistics (ONS) explicitly states that police recorded crime data does not measure crime prevalence and should not be used for this purpose. However, key VAWG policy documents including the Violence Against Women and Girls Strategic Threat and Risk Assessment (STRA), the National Policing Statement on VAWG, and supporting Parliamentary briefings, use police recorded crime figures as direct evidence of epidemic-scale prevalence. This directly contradicts ONS guidance on the appropriate use of its own data.
2. The ONS and OSR Position: Police Recorded Crime Does Not Measure Prevalence
The ONS and the Office for Statistics Regulation (OSR) are explicit and consistent in their guidance on the limitations of police recorded crime data. The following are direct quotes from primary official sources:
a) ONS Domestic Abuse in England and Wales Overview bulletin (November 2025):“Police recorded crime data do not provide a measure of domestic abuse p Caution should be taken when comparing domestic abuse-related police recorded crime data with previous years because of changes in police recording practices.” Source: ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/bulletins/domesticabuseinenglandandwalesoverview/november2025
b) ONS Crime in England and Wales bulletin (year ending March 2024):“Police recorded stalking and harassment should be interpreted with caution, as changes in recording practices and counting rules have affected the recording of violent crime over the last 10 years. Police recorded stalking and harassment does not provide reliable trends in crime but is a better indicator of police activity.” Source: ons.gov.uk/crimeinenglandandwales/yearendingmarch2024
c) ONS Crime in England and Wales bulletin (latest, year ending September 2025):“Police recorded stalking and harassment are not a good indicator of trends over time. This is because changes in recording practices and counting rules have affected the recording of these crimes over the last 10 years.” Source: ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/bulletins/crimeinenglandandwales/yearendingseptember2025#stalking-and-harassment
d) OSR Statement on Crime Statistics (published during the 2024 General Election period):“Many of the claims made this week have been based solely on the police recorded crime statistics…the police recorded crime statistics are a better indicator of police activity than trends in rime…the police recorded crime statistics do not provide reliable long-term trends in crime.” The OSR explicitly directed the public to the CSEW as “a better measure of long term trends in crime.” Source: ostatisticsauthority.gov.uk/news/osr-statement-on-crimestatistics/
e) OSR accreditation status: Police recorded crime statistics lost their accredited official statistics status (formerly National Statistics designation) in 2014. The OSR states: “We removed the accredited official statistics status…in 2014 following an assessment which found evidence that the quality and consistency of the underlying data may not be reliable.” As of the date of this submission, police recorded crime data remains The VAWG STRA and National Policing Statement cited below are built on data that has not held accredited official statistics status for over a decade. Source: osr.statisticsauthority.gov.uk/news/osr-statement-on-crime-statistics/
The ONS and OSR position is unambiguous and has been consistently stated over many years: police recorded crime data measures administrative activity and demand on police services. The CSEW is the appropriate instrument for measuring actual crime prevalence. Importantly, CSEW stalking prevalence data shows the trend has been ‘relatively stable over the last 10 years’ (ONS, year ending March 2025 bulletin), seeming to directly contradict the epidemic narrative in VAWG policy documents.
3. How VAWG Policy Documents Use Police Recorded Crime
The following key policy documents use police recorded crime figures directly as evidence of prevalence, contradicting ONS guidance:
3.1 VAWG Strategic Threat and Risk Assessment (STRA, 2024)
Source: vkpp.org.uk/assets/Uploads/VAWG-Strategic-Threat-and-Risk-Assessment-underpinningand-informing-the-2024-VAWG-Statement-v2.pdf
a) The STRA explicitly states: “VAWG continues to be evidenced at an epidemic scale across England and Wales”, using police recorded crime figures as the primary evidence base for this claim.
b) “The scale of stalking and harassment offences within VAWG remains largely consistent to last year, with 436,196 offences recorded across England and Wales. Stalking and harassment continues to account for the highest proportion of police recorded VAWG crimes (40%).” This recorded crime figure is presented directly as a measure of the scale of the problem.
c) The STRA explicitly served as “the data foundation” for the National Policing Statement and was used to drive responses across government, criminal justice partners, and industry. Its evidence base is wholly police recorded crime data.
3.2 National Policing Statement for VAWG (College of Policing / NPCC, July 2024)
Source: cdn.prgloo.com/media/034ed60aa6564c1fbdcfb03fd8e6a210.pdf
a) The statement presents police recorded figures as evidence of the scale and trajectory of VAWG: “In 2022/23 an estimated 1,080,157 VAWG-related crimes were recorded by the police. This represents a 37% increase in number of VAWG-related crimes compared with 2018/19. This equates to around 2,959 VAWG-related crimes recorded daily and represents just under 20% of all recorded crime in England and Wales.” These figures are presented without any reference to the ONS guidance that police recorded crime does not measure prevalence, and without any reference to CSEW data.
b) The statement cites a 37% increase in police recorded VAWG crime between 2018 and 2023 as evidence that the problem is worsening, with no reference to CSEW prevalence data, which shows stalking trends have been relatively stable over the same period.
3.3 Government and Parliamentary use of these figures
a) The government has committed to halving VAWG within a decade as part of its Safer Streets Mission. The CPS VAWG Strategy 2025-2030 states: “The government has stated its ambition to halve VAWG within a decade through the Safer Streets Mission.” No publicly available document sets out which specific statistical measure will be used to determine whether this target has been met.
b) The government has committed to halving VAWG within a decade as part of its Safer Streets Mission. No publicly available document sets out which specific statistical measure will be used to determine whether this target has been met, a significant accountability gap if the baseline measure is itself unreliable.
4. Why This Matters: Structural Problems with the Data
Beyond the general ONS guidance, there are specific structural reasons why police recorded stalking and harassment figures in particular are unsuitable as prevalence measures:
a) Multiple crime codes per incident: Home Office counting rules mean that a single stalking victim may generate multiple crime codes (8L, 8M, 8Q, 8R) recorded simultaneously. A single course of conduct can produce five or more recorded ‘offences’ in the raw figures.
b) Per-offender recording: A 2016 rule change introduced per-offender recording for certain offences. Where a victim is targeted by multiple people, including in online group harassment scenarios, each alleged offender has the full set of crime codes recorded against them. One victim with five harassers can generate 25 or more recorded ‘offences’.
c) Recording practice changes: The ONS acknowledges that trends in police recorded crime are heavily influenced by changes in recording practices. Theapparent37% rise in VAWG recorded crime between 2018 and 2023 coincides with several significant changes to Home Office counting rules, making year-on-year comparisons unreliable.
d) CSEW data contradicts the recorded crime trend: CSEW prevalence data directly contradicts the epidemic narrative. The ONS states that stalking prevalence “has been relatively stable over the last 10 years” (year ending March 2025 bulletin). This flatly contradicts the 37% increase in recorded crime being used as evidence of a worsening epidemic.
e) Accreditation status: Police recorded crime data lost its accredited official statistics status in 2014 and has not regained it. The ONS domestic abuse bulletin explicitly notes: “Police recorded crime data are not designated as accredited official statistics.” The epidemic narrative underpinning the VAWG national emergency declaration therefore rests on data that does not meet the UK’s own standards for accredited official statistics.
5. The Core Contradiction
The contradiction at the heart of this concern can be stated simply:
a) The ONS states that police recorded crime data does not measure crime prevalence.
b) The ONS recommends the CSEW as the appropriate measure of VAWG prevalence.
c) Key VAWG policy documents, including the STRA, the National Policing Statement, and Parliamentary briefings, use police recorded crime figures as primary evidence of epidemic scale prevalence.
d) The government’s pledge to halve VAWG and the declaration of a ‘national emergency’ in policing appear to rest on this evidence base.
This is not a subtle methodological disagreement. It is a direct and documented contradiction between ONS guidance on the appropriate use of official statistics and their actual use in significant public policy documents. If the ‘epidemic’ narrative and the 37% increase figure have been presented to ministers, Parliament, and the public as evidence of worsening prevalence, this represents a material misuse of official statistics with significant consequences for public policy, resource allocation, and democratic accountability.
6. Questions for the OSR
I would respectfully ask the Office for Statistics Regulation to consider the following:
a) Does the use of police recorded crime data as evidence of VAWG prevalence in the STRA, National Policing Statement, and associated Parliamentary briefings constitute a misuse of official statistics, contrary to ONS guidance and the Code of Practice for Statistics?
b) Is the ONS guidance on the limitations of police recorded crime data for stalking and harassment sufficiently prominent in official publications to prevent its misuse by policy makers?
c) Given that the government has made a public commitment to halve VAWG, and that this commitment is framed around figures the ONS says should not be used as prevalence measures, what steps should be taken to ensure the evidence base for this policy is sound?
d) Should the Home Office, College of Policing, and other bodies using these figures in public communications be required to include appropriate caveats in line with ONS guidance?
7. Note on This Submission
This concern is raised by an independent analyst with no institutional affiliation. It is not motivated by a desire to minimise the serious harm caused by violence against women and girls, which is a genuine and important public concern. The concern is solely about the integrity of the statistical evidence used to underpin public policy, and whether that evidence is being used in ways consistent with the guidance of the national statistical authority.
The documents referenced in this submission are all publicly available official publications. I am happy to provide further detail or clarification on any aspect of this concern if that would assist the OSR’s consideration.
In preparing this submission I noticed that the OSR issues log records a prior case (2025/098: VAWG measurement, raised 14 November 2025, formal response issued privately 15 January 2026). As no further detail is publicly available on that case, I cannot know whether my concern overlaps with, duplicates, or is entirely distinct from it. I raise this simply in the interest of transparency, there is no implication of any difficulty with that prior case. I would simply ask that if my concern has already been substantively addressed in case 2025/098, the OSR let me know so that I can consider whether any further action is warranted in light of that outcome.
Email correspondence from OSR (25 March 2026 )
Thank you for raising yours concerns about the use and presentation of police recorded crime statistics in the national policing statement on Violence Against Women and Girls (VAWG) and the Strategic Threat and Risk Assessment (STRA).
You asked us to consider four questions. We address each in turn below.
Misuse of official statistics
Police recorded crime is the key data source for the statement and STRA. We would expect national policing bodies to draw heavily on police recorded crime data, as well as related data sources like the Crime Survey for England and Wales (CSEW).
While the information about limitations of police recorded crime data in both documents could be enhanced, we do not consider that the presentation of the statistics in the statement and STRA constitute misuse.
The police recorded crime statistics are used to support the narrative that scale of VAWG is an epidemic, but both documents acknowledge that many VAWG-related crimes are not reported to the police. In addition, the statement includes a chart comparing figures on the estimated on the estimated number of victims of domestic abuse and sexual assault from the CSEW and police recorded crime, to highlight the ‘vast and largely hidden extent of VAWG’. This is helpful supporting information, although the chart is not well-explained or contextualised and therefore may not be easily understood by users with a limited knowledge of the CSEW.
It is important to make a distinction between the terms ‘prevalence’ and ‘scale’. As you are aware, prevalence measures the proportion of the population that has been a victim of crime while scale refers to the number of crimes that have occurred. The statement and STRA largely discuss the scale and nature of VAWG, not prevalence. The statement does include two prevalence estimates (of VAWG victimisation and VAWG perpetration) but these come from the CSEW and an academic survey, respectively.
ONS’s guidance on limitations
In our view, ONS’s guidance on the limitations of police recorded crime data is sufficiently clear and prominent. For instance, the latest Crime in England and Wales bulletin includes the prominent caveat that: ‘Police recorded stalking and harassment are not a good indicator of trends over time. This is because changes in recording practices and counting rules have affected the recording of these crimes over the last 10 years.’
VAWG strategy and metrics
The UK Government published its VAWG strategy in December. It sets out the monitoring framework that the Home Office is using to measure its ambition to halve VAWG within a decade (on pages 69-71).
The Crime Survey for England and Wales (CSEW) underpins the monitoring framework; most of the headline metrics are based on the CSEW, not police recorded crime. This includes the key headline metric, which is a combined measure of the prevalence domestic abuse, sexual assault and stalking among people aged 16 and over. ONS published the first estimate of the combined prevalence measure in 2025.
The other VAWG case mentioned in your submission focuses on the quality and suitability of ONS’s combined prevalence measure. As a result, there is only limited overlap with the concerns you have raised.
Only the ‘female homicides’ supporting headline metric is based on police recorded crime data. It appears that this is based on the Homicide Index, which is the only component of the police recorded crime statistics that is published as accredited official statistics.
It is good that the CSEW is the primary data source for monitoring progress against the VAWG ambition, as it is the most reliable source of information on VAWG-related crimes.
To support public understanding of the monitoring framework, we think the Home Office should publish more detailed information about its choice of metrics and the data sources for the metrics.
Caveats in public communications
We agree that appropriate caveats should be presented alongside police recorded crime statistics, to ensure accurate interpretation of the figures.
The STRA prominently highlights several data limitations, including underreporting to the police, inconsistencies in IT crime recording systems across police forces, and data coverage issues and gaps. For instance, the section on stalking and harassment states that ‘it is almost certain that the majority of stalking and harassment offences online go unreported by women and girls to police’.
The statement also mentions some limitations and challenges of measuring VAWG, including that VAWG is a collection of offences, not a single crime. And there is the comparison between the CSEW and police recorded crime figures which we highlighted earlier.
The key limitation of the police recorded crime data that is not covered by either the statement or the STRA is that police recorded crime figures are a poor indicator of trends in VAWG-related crime over time due to changes in police recording practices. We think that changes to police recording practices should be acknowledged as this affects interpretation of the statistics. As you rightly point out, changes in recording practices have affected some crime types more than others, and this can make year-on-year comparisons unreliable.
ONS’s guidance states that police recorded crime figures are best used to understand police workload and demand. The statement does at least describe the 37% increase in VAWG-related crimes between 2018/19 and 2022/23 in the context of the increased scale of demand on policing (‘the scale of demand on policing from VAWG-related crime is increasing’). But this is not done consistently throughout the document and the STRA only mentions demand once.
As we are generally satisfied with the presentation of the police recorded crime statistics in the statement and STRA, we are not going to ask the College of Policing and NPCC to update these documents. However, we will monitor future uses of the police recorded crime statistics related to VAWG, and where possible, encourage clear communication of the limitations of police recorded crime data.
Yours sincerely,
Ed Humpherson, Head of the Office for Statistics Regulation
Email correspondence to OSR (27 March 2026)
Dear OSR team,
Thank you for your detailed response. I appreciate the time taken to address each of the four questions I raised.
Having reviewed your answers, there is one point on which I would be grateful for clarification.
In response to Question (a), you state that the STRA and National Policing Statement “largely discuss the scale and nature of VAWG, not prevalence,” and therefore their use of police recorded crime does not constitute misuse. However, both documents use police recorded crime figures in ways that appear to describe or imply worsening prevalence; for example, the STRA’s reference to VAWG being “evidenced at an epidemic scale” and the National Policing Statement’s presentation of a 37% increase in VAWG‑related recorded crime between 2018/19 and 2022/23.
At the same time, ONS guidance states that police recorded crime data “do not provide a measure of domestic abuse prevalence” and that police recorded stalking and harassment figures “are not a good indicator of trends over time” due to changes in recording practices.
To ensure I have correctly understood your position, could you please clarify the following:
Is it acceptable under the Code of Practice for policy documents to use police recorded crime figures in ways that describe or imply worsening prevalence (for example, “epidemic scale” or citing percentage increases over time), given ONS’s explicit guidance that these figures do not measure prevalence and are a poor indicator of trends?
I am not seeking to reopen the wider issues raised in my submission; I would simply appreciate clarity on this specific point so that I can accurately interpret your response.
Thank you again for your time.
Email correspondence from OSR (29 April 2026)
Thank you for your further email and apologies for the delay in responding.
You asked whether it is acceptable, under the Code of Practice for Statistics, for policy documents to use police recorded crime figures in ways that describe or imply worsening prevalence, given the Office for National Statistics (ONS)’s guidance that these figures do not measure prevalence and are a poor indicator of trends for many crime types.
Standard 2 of the Code’s Standards for the Public Use, Data and Wider Analysis states that ‘Public bodies should use statistics, data and wider analysis with integrity and communicate them with clarity and accuracy, so that the public can easily understand the basis for claims and decisions made’. In practice, this means that statistics should not be presented in a misleading way and that any key context and limitations associated with the statistics should be made clear to support appropriate interpretation.
Our view is that the police recorded crime statistics included in the national policing statement STRA are not, in themselves, used in a misleading way. However, we consider that the context provided alongside the statistics could be expanded, particularly in the STRA, and that some key limitations are not communicated as clearly or consistently as they could be.
It is appropriate to discuss changes in the number of VAWG-related offences recorded by police over time when this is framed in the context of demand on the police. This is the case for the policing statement, which explicitly links increases in recorded crime to the scale of demand on policing, but this contextual framing is missing from the STRA.
You specifically highlighted the use of term ‘epidemic’. The word appears only once in each document, but we recognise that it carries strong connotations and that it has been used widely since the policing statement was published to discuss the scale of VAWG. In the introduction to the STRA, the sentence about VAWG being evidenced at an epidemic scale follows the sentence about the scale of VAWG-related crime recorded by the police, but they are separate. We note that the main source of evidence used to support the claim that VAWG is at an epidemic scale in the STRA is the widespread reporting of serious VAWG-related incidents in the media. In addition, the introduction to the STRA does not cite the percentage change in VAWG-related offences recorded by the police over time.
We discussed your concern with the Office for National Statistics (ONS). ONS statisticians agreed with our assessment that the figures in the executive summary of the policing statement are appropriately contextualised, that the reporting in the STRA could be improved, and that the presentation of the statistics is not misleading.
Police recorded crime statistics on VAWG are frequently presented by a range of users, including the media and other public bodies, as an indicator of trends, despite ONS’s clear guidance and prominent caveats. In many cases, this does not amount to misuse but rather reflects insufficient contextualisation.
I hope this clarifies our view. We will pass on your feedback to the Home Office and NPCC.
Yours sincerely,
Ed Humpherson, Head of the Office for Statistics Regulation