Redevelopment of domestic abuse statistics: Research update November 2025

The way domestic abuse manifests is constantly changing. Because of this, as well as legislative changes, we have undertaken a user engagement, research and testing programme to ensure our statistics continue to provide the most accurate information and meet the needs of users. Through this work, we developed a new set of survey questions for the Crime Survey for England and Wales (CSEW), which have now been accredited as the official statistics on domestic abuse from the year ending March 2024 onwards. More information on the new questions can be found in our Evaluating a new measure of domestic abuse article.

The new questions provide the opportunity to produce measures of the scale of domestic abuse that victims experienced. The scales account for both the abusive behaviours experienced and the impacts of the abuse and will show the spectrum of victims based on their experiences since the age of 16 years. This is something that is not currently available in our Domestic abuse in England and Wales compendium. The measures aim to help policy makers and victim services in understanding types of victims and developing policies and services to best meet their needs.

This article presents abuse scales for the first time, including the method behind them and resulting estimates. We will work with users to continue to develop these scales and provide the support needed to understand the estimates.

The new domestic abuse questions were designed so they could be used to derive measures based on abusive behaviours and the impacts of those behaviours.

The survey questions (see Section 16 of our Crime Survey for England and Wales 2025 to 2026: Adult Questionnaire, PDF, 3.8MB) ask respondents whether they have experienced abusive behaviours by a partner or family member since the age of 16 years. If the respondent states they have experienced any of the behaviours, they are presented with a set of impacts and asked if they have experienced these as a result of the behaviours since the age of 16 years.

The questions also ask respondents whether they have experienced abusive behaviours in the last year but do not ask questions on impact for the same time period. This means it is not possible to derive abuse scales for the "in the last year" time period. Instead, the abuse scales will measure abuse experienced by victims since the age of 16 years.

As such, the abuse scales will not differentiate between when the abusive behaviours occurred and when the impacts of those behaviours were felt. It also means it is not possible to identify if behaviours and their impacts are recent or date back many years.

More information on the structure of the new domestic abuse questions can be found in Section 4.2.2 of our Evaluating a new measure of domestic abuse article.

Our aim was to produce a measure that accounted for the scale of domestic abuse that victims had experienced since the age of 16 years. We did this by considering both the number of different abusive behaviours victims had experienced as well as the number of different impacts of the abuse.

We already publish headline prevalence measures of domestic abuse in our Domestic abuse in England and Wales compendium. These measures are based on experiencing one or more abusive behaviours. However, as victims may have experienced one or many abusive behaviours, and the impacts felt will vary greatly, analysis of abuse scales provides important insight to consider alongside the headline measures.

The measures presented in this article were referred to in previous publications as abuse profiles, however the term abuse scales more accurately describes the measure we have developed.

By analysing the number of abusive behaviours and the number of impacts victims have experienced, we are able to group victims with similar experiences. This is done using the statistical method of clustering. The resulting abuse scales can then be analysed to understand the different experiences of victims.

Using the behaviour and impact questions, we created variables summing the number of different abusive behaviours a respondent had experienced since the age of 16 years, and the number of resulting impacts they selected. We did this separately for partner abuse, family abuse and overall domestic abuse. These variables were used to conduct our analysis.

While the abuse scales account for the number of different abusive behaviours a victim experienced, the frequency of experiencing the same behaviour multiple times is not measured by the survey, apart from for physical abuse, and is therefore not accounted for. The scales do not make any assessment or comparison of the severity of behaviours or impacts as this is subjective and will be different for each victim. More information on our domestic abuse questions can be found in our Crime Survey for England and Wales 2024 to 2025: Adult Questionnaire (PDF, 3.2MB).

We explored the use of a variety of methods to help us understand the scale of domestic abuse experienced by victims. We collaborated with academics from the University of Southampton to best utilise the data we had available from the new domestic abuse questions. We decided to use a clustering methodology to identify groups of victims who had experienced similar numbers of different abusive behaviours and impacts. This method allowed us to flexibly identify patterns in the data.

The method of clustering used was k-means. The k-means algorithm takes a chosen number of random centroid points (denoted by k) and assigns each individual data point to its closest centroid, resulting in the formation of k initial clusters. The total distance between points and their respective centroid is calculated and stored. The algorithm then calculates new centroid points as the centre of each cluster. The points are again assigned to their nearest centroid, forming new clusters, and the total distances to the centroid within each cluster is recalculated and stored. This process is repeated until the centroid points no longer change and the total distance between points and their respective centroids is minimised.

We performed our cluster analysis on the data collected from the new survey questions between April 2023 and March 2025. The sample size for this period is smaller because it covers the period when we were trialling the new domestic abuse questions. Therefore, the uncertainty of the results is higher than if the questions had been asked to a full sample. Our analysis showed that for domestic, partner and family abuse, the k-means algorithm produced three victim clusters. For domestic abuse and partner abuse, the clusters are defined as:

• Cluster 1: Lower numbers of abusive behaviours and impacts

• Cluster 2: Medium numbers of abusive behaviours and impacts

• Cluster 3: Higher numbers of abusive behaviours and impacts

Although the clusters for domestic abuse and partner abuse have the same labels, they are not comparable. This is because the domestic abuse clusters include both partner abuse and family abuse victims.

For family abuse, the patterns of behaviours and impacts differed compared with domestic abuse and partner abuse – our analysis showed that the relationship between behaviours and impacts was more positive and linear. The differences were largely seen in Cluster 2, where family abuse victims selected lower numbers of behaviours but higher numbers of impacts.

For family abuse, the clusters are defined as:

• Cluster 1: Lower numbers of abusive behaviours and impacts

• Cluster 2: Lower numbers of abusive behaviours and higher numbers of impacts

• Cluster 3: Higher numbers of abusive behaviours and impacts

We have currently used the term "clusters" when labelling the different groups, as we believe there is further work to be done to develop accurate labelling for the different clusters. We welcome feedback on this. Please email your comments to: crimestatistics@ons.gov.uk.

A victim may be classified as experiencing low numbers of abusive behaviours but could have experienced one abusive behaviour many times. In addition, victims of domestic abuse may experience similar behaviours, yet the impacts felt from those behaviours may differ greatly. It is important to note that, irrespective of cluster, all are victims of domestic abuse in accordance with the definition in the Domestic Abuse Act 2021 and will be counted in our headline prevalence measures.

While the production of abuse scales provides additional information on victims of domestic abuse, there are some limitations to this method. For example, the optimal number of clusters may change over time, which will affect comparability. Although year-to-year comparisons of lifetime measures generally only show sampling variation, over longer periods changes can be seen, therefore comparability is important. As we collect more data from the new questions, we can monitor trends in the data and the number of clusters. If this differs from three, there is the possibility of re-running previous years' data to match the new number of clusters, which would allow for time series comparisons. We will be working with users to understand their requirements for comparisons of the scales over time and how they would like to use these data before finalising our methods.

Following the creation of our abuse scales containing three groups, we have produced three different measures for domestic abuse, partner abuse and family abuse showing:

• the percentage of people aged 16 years and over by abuse scale since the age of 16 years

• the percentage of victims aged 16 years and over by abuse scale since the age of 16 years

• the percentage of people aged 16 years and over that were victims of abuse in the last year by abuse scale since the age of 16 years

Statistics presented in this section are official statistics in development and are subject to change. They are presented here to aid users in providing feedback while we evaluate and finalise our abuse scales methodology.

Data are presented for year ending March 2025. Data for year ending March 2024, along with full breakdowns for both years, are published in our Redevelopment of domestic abuse statistics, abuse scales since the age of 16 years, England and Wales data tables. Comparisons between the two years have not been made in this article as differences are likely down to sampling variation.

Prevalence of domestic abuse since the age of 16 years by abuse scales

The Crime Survey for England and Wales (CSEW) year ending March 2025 estimated that 25.8% of people aged 16 years and over have experienced domestic abuse since the age of 16 years. Abuse scales can be used to show how domestic abuse experiences vary within our overall prevalence estimate.

The percentage of people who were victims (prevalence rate) of domestic abuse, partner abuse and family abuse since the age of 16 years was highest for Cluster 1, where victims experienced lower numbers of abusive behaviours and impacts. Prevalence rates were lowest for victims in Cluster 3, experiencing higher numbers of behaviours and impacts.

Around 8.4 million people aged 16 years and over (17.2%) have experienced lower numbers of domestic abuse behaviours and impacts since the age of 16 years. This compares with 3.2 million people (6.7%) who have experienced medium numbers of domestic abuse behaviours and impacts, and 891,000 people (1.8%) who have experienced higher numbers of domestic abuse behaviours and impacts (Figure 1).

The prevalence of experiencing medium and higher numbers of abusive behaviours and impacts since the age of 16 years (Clusters 2 and 3) was significantly higher for female victims of domestic abuse compared with male victims (Figure 2). The same pattern was seen for experiences of partner abuse.

For experiences of family abuse since the age of 16 years, females in Cluster 2 had significantly higher prevalence rates compared with males. More information can be found in our Redevelopment of domestic abuse statistics, abuse scales since the age of 16 years, England and Wales data tables.

Percentage of domestic abuse victims by abuse scales

An alternative way to present the data on abuse scales is to analyse the percentage of domestic abuse victims who fell into each of the three clusters.

The majority of victims of domestic abuse since the age of 16 years were in Cluster 1, where victims experienced lower numbers of abusive behaviours and impacts. The percentage of victims fell as the numbers of abusive behaviours and impacts increased. This relationship was seen for domestic, partner and family abuse (Figure 3).

The proportion of males and females in Cluster 1 was higher than in Clusters 2 and 3, where the numbers of domestic abuse behaviours and impacts were higher. Proportionally there were more females in Clusters 2 and 3 than males (Figure 4).

Prevalence of domestic abuse in the last year by abuse scales

The new domestic abuse questions ask victims whether they have experienced domestic abuse in the last year. However, they do not specifically ask whether impacts were felt as a result of that abuse in the last year. This means we cannot produce estimates showing abuse scales based on abusive behaviours and impacts experienced in the last year.

Instead, for those victims who experienced at least one abusive behaviour in the last year, we can analyse their abuse scale since the age of 16 years. The impacts used to create these scales may not have been felt in the last year and could be the result of abuse experienced over 12 months ago. Therefore, this is not a proxy measure for abuse scales in the last year. However, analysing these data provides a way of understanding the types of victims that have experienced recent abuse.

For the year ending March 2025, the CSEW estimated that 7.8% of people aged 16 years and over experienced domestic abuse in the last year.

Broken down by abuse scale since the age of 16 years, 4.2% of people who had experienced an abusive behaviour in the last year had experienced low numbers of domestic abusive behaviours and impacts since the age of 16 years (Cluster 1). The prevalence of people who had experienced domestic abuse in the last year and experienced medium and higher numbers of abusive behaviour and impacts since the age of 16 years was 2.7% and 0.9%, respectively (Figure 5).

A higher proportion of females who had experienced domestic abuse in the last year were in Clusters 2 and 3 (meaning that they had experienced medium and higher numbers of abusive behaviour and impacts since the age of 16 years) compared with males. Breakdowns by partner and family abuse can be found in our Redevelopment of domestic abuse statistics, abuse scales since the age of 16 years, England and Wales data tables.

When looking at the percentage of victims in the last year by abuse scale since the age of 16 years, a similar pattern to that presented in Figure 3 was seen. More information on these estimates can be found in Tables 5a and 5b of our Redevelopment of domestic abuse statistics, abuse scales since the age of 16 years, England and Wales data tables.

Statistics presented in this article are official statistics in development and are subject to change. We have shared our methodology and estimates with the aim of collecting feedback as we look to finalise our method for producing abuse scales.

Users can contact us with comments and feedback on the data presented in this article and accompanying tables at crimestatistics@ons.gov.uk. We are particularly interested in hearing views and feedback on:

• which of the abuse scale estimates are most useful – estimates as a percentage of the entire population or as a percentage of victims

• what additional estimates of abuse scales would be useful

• the naming and descriptions of the clusters

• the presentation of the data tables

This feedback will help us understand what estimates we should focus on producing and enable us to provide further breakdowns to meet user needs where possible.

Once our methodology has been finalised, we aim to publish the abuse scales alongside our next annual domestic abuse compendium.