How a change in the Crime Survey for England and Wales questionnaire in April 2025 affected responses to questions on anti-social behaviour

Data from the Crime Survey for England and Wales (CSEW) showed a notable rise in people experiencing or witnessing anti-social behaviour (ASB) in the local area for year ending March 2026.

This supporting methodology article presents findings from our analysis to investigate whether increases in ASB were genuine or because of changes to how questions in the CSEW were asked from April 2025. CSEW respondents are randomly allocated to four different pathways in the survey (Routes A, B, C, and D). The main changes were the removal of questions on household security in Route B and on vehicle security in Route C.

Results showed that there were increases in experiencing or witnessing ASB across all four pathways but that the increase was larger in Routes B and C, where questions on household and vehicle security were removed. This suggests most of this increase was genuine, but that it was inflated because of these changes to how questions in the CSEW were asked from April 2025.

The Crime Survey for England and Wales (CSEW) is a face-to-face survey in which a nationally representative sample of the population aged 16 years and over who are resident in households in England and Wales are interviewed in their own home.

All CSEW respondents answer questions on perceptions and experience of crime at the beginning of the survey. To enable consistency in the time series of our headline crime estimates, there has been minimal change to these questions over time. However, changes and improvements are often made to subsequent areas of the survey by adding, removing or amending questions when necessary.

After these initial questions, respondents are generally asked about the following topics:

• performance of the criminal justice system

• mobile phone crime

• experiences of the police

• crime prevention and security

• anti-social behaviour (ASB)

• harassment

• drugs and drinking

• domestic abuse, sexual victimisation and stalking

Respondents are randomly allocated to one of four pathways (Routes A, B, C, or D) and they answer different sets of questions in some parts of the survey. This process is used to maximise the number of questions that we can ask, while ensuring that each split-sample module remains representative of the survey population. For example, Route A respondents in the year ending (YE) March 2025 survey were not asked questions on crime prevention and security and were asked additional questions on perception of ASB in the ASB module. However, Route B respondents were asked about household security, Route C respondents were asked about vehicle security, and Route D respondents were asked about personal and online security. Please see our CSEW questionnaires for exact routing of questions by each survey year.

The CSEW has measured experiencing and witnessing anti-social behaviour (ASB) since 2012, and the wording of the questions and response options have remained the same. They are:

• "Have you personally experienced or witnessed any sort of anti-social behaviour in your local area in the last 12 months?"

• "What sorts of anti-social behaviour have you personally experienced or witnessed in your local area in the last 12 months?"

Data on ASB are published in our Crime in England and Wales: appendix tables and our Crime in England and Wales: annual supplementary tables.

April 2025 changes to survey questionnaire

In April 2025, changes were made to the crime prevention and security section of the survey, which affected Routes B and C. Questions on household and vehicle security were removed to make space for other questions later in the survey. This changed the set of questions that were asked of respondents in Routes B and C directly before they answered questions on ASB. Minimal changes were made to Routes A and D in this part of the survey.

Research literature on survey methodology has shown that the context and order in which questions appear in a survey can influence the answers given by respondents. Such effects tend to be less striking for questions about facts (such as age or employment status) than for questions about attitudes and experiences, which are more sensitive to the context and order in which questions are asked.

A change to the order of questions in the CSEW in April 2018 was previously shown to only have had a small effect on our estimates of experiencing or witnessing ASB. For more information, see our Methodology note on the possible order effect on responses to questions on anti-social behaviour from changes in the CSEW questionnaire.

The following sections describe our methodology and the results of our evaluation of how the April 2025 CSEW questionnaire updates affected our ASB estimates.

We analysed responses to experiencing or witnessing anti-social behaviour (ASB) questions using data collected in the Crime Survey for England and Wales between April 2023 and March 2026.

Prevalence estimates for experiencing or witnessing ASB were calculated for year ending (YE) March 2025 and YE March 2026, as well as for each month between April 2023 and March 2026. Prevalence estimates were also calculated for experiencing or witnessing ASB by the four respondent pathways (Routes A, B, C, or D). Significance testing was also conducted to test if changes between the two survey years for the whole sample and the four respondent pathways were statistically significant.

To identify if changes in estimates were because of questionnaire changes or a genuine change in trend, we used interrupted time series analysis. This method is a form of regression analysis, where time is partitioned to estimate the impact of an intervention while also accounting for the general trend pre- and post-intervention. In this analysis, the intervention was represented by the April 2025 questionnaire changes and estimates were divided into two time periods: pre-intervention and post-intervention.

In the adjusted regression model, we calculated coefficients for the two time periods to account for the existing trend in the data when estimating the intervention effects. We then calculated odds ratios for the association between experiencing or witnessing ASB, and intervention effects. For definitions of regression analysis and odds ratios, see Section 5: Glossary.

Experience of any anti-social behaviour

Crime Survey for England and Wales (CSEW) estimates of experience of any anti-social behaviour (ASB) in the local area from the year ending (YE) March 2019 survey to the YE March 2025 survey have fluctuated between 34% and 40% (Table C1 in our accompanying dataset). However, there was a large, one-year increase in experiencing or witnessing ASB from the YE March 2025 to the YE March 2026 survey (35.5% to 40.9%).

Figure 1 provides the prevalence estimates for experiencing or witnessing ASB from April 2023 to March 2026 using monthly data (Table 2 in our accompanying dataset). Although there is a step change in April 2025, there has also been a gradual increase since April 2025, suggesting some of the rise may be genuine.

Figure 1: Proportion of people who experienced or witnessed anti-social behaviour in the last 12 months

Experiences of or witnessing anti-social behaviour, by month of interview, Crime Survey for England and Wales, April 2023 to March 2026

Results from the interrupted time series analysis showed a significant association between the intervention and an increase in experiencing or witnessing ASB, after accounting for trends pre- and post-intervention (odds ratio 1.18; Table 3 in our accompanying dataset). This suggests that the questionnaire changes had partially contributed to the increase. For the definition of odds ratio (OR), see Section 5: Glossary.

Experience of any anti-social behaviour by respondent pathway

CSEW estimates of experiencing or witnessing ASB showed increases in all groups, when split by respondent pathway in the YE March 2026 survey, compared with the YE March 2025 survey (Table 1 in our accompanying dataset). However, there were differences between these groups. While Routes A and D saw a 4 to 5 percentage point increase in experiencing or witnessing ASB, there was a 6 to 7 percentage point increase for Route B and C respondents.

Figure 2 shows the monthly prevalence estimates of experiencing or witnessing ASB from April 2023 to March 2026 for the four split-sample modules (Table 2 in our accompanying dataset). Most notably, there was a distinct increase in experiencing or witnessing ASB following questionnaire changes for Route B and C respondents. However, there were no obvious step changes in Routes A or D.

Figure 2: Proportion of people who experienced or witnessed anti-social behaviour in the last 12 months

Experiences of or witnessing anti-social behaviour, by respondent pathway and month of interview, Crime Survey for England and Wales, April 2023 to March 2026

Our adjusted regression analysis showed that increased experienced or witnessed ASB was only associated with the intervention for those in Route B (OR 1.23; Table 3 in our accompanying dataset) and Route C (OR 1.32; Table 3 in our accompanying dataset). There was no association between increased experienced or witnessed ASB and questionnaire changes for the other respondent pathways (Routes A and D). For Route A, time after the intervention was associated with increased ASB in fully adjusted models. This suggests that the questionnaire changes have also been accompanied by a genuine rise in experienced or witnessed ASB.

Although the intervention in Routes B and C was significantly associated with increased experienced or witnessed ASB, the effect was fairly small and only partly affected prevalence estimates. In YE March 2026, 40.9% of respondents reported experienced or witnessed ASB in the total sample. This is a 5.4% increase, compared with YE March 2025 (35.5%). Experienced or witnessed ASB still increased by 4.4% in YE March 2026 if Route B and C were excluded, suggesting that most of the increase was genuine (4.4% out of the 5.4% increase).

The association between the questionnaire changes in Routes B and C with increased experienced or witnessed ASB could be the result of cognitive-based effects. Respondents in Routes B and C were asked questions related to household and vehicle security, respectively, in the YE March 2025 survey. It is possible that these questions on physical security may have influenced their thoughts when answering questions on experiencing or witnessing ASB in the local area. Therefore, the omission of these questions may have affected how they answered questions on experienced or witnessed ASB in the YE March 2026 survey. Their omission also meant that questions related to experiencing or witnessing ASB were now closer to questions on perceptions of the police. This may have also influenced respondents' thoughts when answering questions on experiencing or witnessing ASB.

Conclusions and recommendations

Although most of this increase in experienced or witnessed ASB was genuine, it was inflated because of changes to how questions in the CSEW were asked from April 2025. Given that the impact of the questionnaire changes was fairly small, we will continue to base our estimates on the full sample.

As the nature of crime continues to change, the CSEW questionnaire must evolve to capture emerging issues and meet user needs. Given the limited space available on the survey, this often requires respondents to be routed through different questionnaire pathways and for some content to be rotated in and out over time.

We will continue to review any planned changes to the CSEW questionnaire to reduce the risk of similar impacts in the future. This includes monitoring additional estimates by quarter of interview, so that we can identify any concerns at an earlier stage.

Regression analysis

A statistical method used to examine the association between a specific characteristic and an outcome. It determines if a predictor variable is strongly linked to a higher or lower likelihood of an outcome. This analysis quantifies how much of the difference in estimates between groups can be explained by a predictor, while often "controlling" for other factors to ensure the relationship is direct.

Odds ratio

An odds ratio for a particular group describes the relative difference in the likelihood of an outcome in that group compared with a reference category, which was based on majority count in this analysis. An odds ratio higher than 1 indicates a greater likelihood of that outcome, while an odds ratio less than 1 indicates a lower likelihood.

Confidence intervals

Confidence intervals use the standard error to derive a range in which we think the true value is likely to lie. A confidence interval gives an indication of the degree of uncertainty of an estimate and helps to decide how precise a sample estimate is. It specifies a range of values likely to contain the unknown population value.

These values are defined by lower and upper limits. The width of the interval depends on the precision of the estimate and the confidence level used. A greater standard error will result in a wider interval; the wider the interval, the less precise the estimate.