How the Crown Prosecution Service is understanding and tackling racial disproportionality in our work
The Crown Prosecution Service sits at the heart of the criminal justice system. Our prosecutors decide whether someone should be charged with the most serious criminal offences. The decisions we make have a profound impact on the lives of victims and defendants and it is imperative that we hold ourselves to the highest standards of accountability and transparency.
This is why we carried out a comprehensive programme of research to scrutinise our own work and hold ourselves to the highest standards.
Through this research we have identified that there has been evidence of disproportionality in our decision making. We’ve also identified ways we can address the root causes of this – set out in our rigorous action plan. Our research findings have been at the centre of developing our action plan – making sure we can bring meaningful change that will target any disproportionality effectively and robustly.
Phase One
In 2021, we commissioned independent research from the University of Leeds to examine the outcomes of our decision making by sex, age and ethnicity.
NB: This research is out of date. Read more about the latest figures below, under 'Disproportionality: Updated 2022-2024 Research'.
Phase Two
To better understand what was driving this disparity and what actions we could take to tackle this, we wanted to look our cases and processes in more depth.
In 2023, we began this work with scrutiny from an independent Disproportionality Advisory Group comprised of academics, legal professionals, and third-sector partners.
This stage looked at how our decisions may be impacted by things like previous offending history, the language used in evidence files and the processes we follow when reviewing them.
We used a variety of research methods in this phase. We looked at a sample of 400 like-for-like cases of white British and mixed ethnicity suspects (as these groups had the highest disparity in our original research.) We also carried out interviews and a survey with our staff.
Overall, the research found that the underlying causes are complex, but reinforced that disproportionality existed.
Our research was made up of a series of research questions – each aimed to test a possible reason for disproportionality. The findings are set out below:
Suspect and case characteristics
Could certain case or suspect characteristics explain the disproportionality? For example was disproportionality caused by previous convictions or socio-economic status, rather than race?
The research found no significant differences in characteristics that could have explained the disparity. This included that previous convictions and socio-economic status were not a factor.
We did find differences in the age of the suspects, with mixed ethnicity suspects being younger than white British suspects. We also found that white suspects were expected to plead guilty more often than mixed ethnicity suspects.
Information provided in case files
Does the amount of information and evidence provided by the police and prosecutors vary depending on suspect ethnicity? As this could be influencing the final charging decision.
We looked at the word count on documents provided by the police to the CPS as well as the number of evidential files passed on to us – there was no difference in the amount of information provided.
We also looked at the word count of our charging decisions, which is the information CPS prosecutors provide when reviewing the case. On average, there was a higher word count provided for mixed ethnicity suspects than white British suspects.
Language used in case files
Is the language used to describe suspects and cases different for white suspects and mixed ethnicity suspects? As this could also be influencing the final charging decision.
We commissioned Aston University to carry out this piece of research. They analysed large samples of text to uncover implicit patterns in the language used. They compared like-for-like cases with similar crime types and levels of harm.
The study found a clear difference in the language used by both police and prosecutors. We did not find explicit racism or biases in the texts, however more direct and definitive language was used in respect of mixed ethnicity suspects, their actions were described in more negative terms and more references to the violent nature of their offences.
Working practices
Do the CPS’ practices and processes contribute towards disproportionality in decision making?
For this part of the research, we conducted a series of surveys and interviews with staff around equality and diversity initiatives, legal training and how the Code for Crown Prosecutors is applied.
Research with staff identified:
- Workloads were seen as a barrier for staff to prioritise equality and diversity initiatives or challenge unconscious bias
- Senior leadership will be pivotal in championing equality and diversity, and driving change
- CPS legal training supports prosecutors in making fair decisions, however more could be done to warn against making disproportionate decisions
- Research on how the Code for Crown Prosecutors is applied revealed the potential for unconscious bias to have an impact.
Download and read the full report (PDF document, 372kb). This document may not be suitable for users of assistive technology.
Our action plan
Our action plan - developed collaboratively with colleagues and external partners - will enable us to develop an anti-racist culture and practices, eliminating racial bias in our decision-making and working with other criminal justice partners to address race disproportionality across the system.
We will carry out a comprehensive review of the Code for Crown Prosecutors – the guidance prosecutors follow when making legal decisions - and amend it to ensure that identifying and addressing bias is considered by prosecutors in all their casework decisions.
We will develop extensive learning and development, training and new digital tools for our staff, and specifically prosecutors.
We will also prioritise working closely with our partners across the criminal justice system. This includes establishing a Joint CPS and National Police Chiefs' Council Race Disproportionality Board, which allows for continuous check and challenge between prosecutors and police.
We will identify ways of further strengthening our stakeholder and community engagement channels, including involving experts in addressing disproportionality.
The diversity of our workforce – one of the most diverse in the Civil Service and criminal justice system – is one of our strengths and the plan will help us to build on this further. We will focus on improving recruitment, retention and development of ethnic minority staff to drive better representation of the communities we serve.
Read the full action plan
We are delivering on our Disproportionality Action Plan. A focus has been on ensuring we have the latest evidence base from which to build our work, so we have replicated the original analysis looking at charging decisions using a more up-to-date time period. We corrected an issue with this analysis. The updated findings of disproportionality levels are available below.
Our research team has also worked with the Behavioural Insights Team to understand the most effective ways to tackle racial disproportionality and strengthen our anti-racist culture. We will embed the findings into our work as we continue to progress.
We have published our Anti-racist statement, a first report on the Joint Enterprise Monitoring Scheme and consulted on our ‘gang’-related prosecution guidance.
We have set up the CPS and National Police Chiefs' Council Joint Race Disproportionality Board, recognising that tackling disproportionality cannot be done in isolation and requires a whole system approach across the criminal justice landscape. The board provides an opportunity to develop a shared understanding and culture of anti-racism, drive collaboration and improve data collection to enable better monitoring.
As part of a wider programme of learning to support prosecutors in identifying and addressing bias, we have launched the first of four practical, case study-led resource packs focused on disproportionality and enhancing the way prosecutors build cases.
In August 2025, we hosted interns across England and Wales as part of the Bar Council’s 10,000 Black Interns scheme. We will continue to support this important initiative to help talented young people from underrepresented backgrounds access careers in the legal sector.
Introduction
As part of the Disproportionality Action Plan to continue to monitor disproportionality, the Crown Prosecution Service research team analysed data from 2022-2024 to bring the findings up-to-date, using the methods from the original analysis. The latest research has also corrected an issue with the original analysis which had under-counted the number of defendants, leading to levels of disparity being over-stated.
Background
Disproportionality analyses to date have covered two time periods:
- 2018-2021 data – covered by original analysis carried out by University of Leeds academics, the results of which were published;
- 2022-2024 data – this analysis was carried out by the CPS’s in-house Research Team, who ran the analysis on an updated time period.
The data used in the original analysis was in an aggregated format rather than being disaggregated at an individual level; however, the original analysis was conducted on the basis that the data was disaggregated. This caused an undercount of suspects. The 2018-2022 analysis has now been re-run using disaggregated data.
The analysis looked at charging decisions for principal offenders for the two time periods that led to a charge, no further action (NFA) or a caution – amounting to nearly 420,000 cases in 2018-2022 and 470,000 cases in 2022-24, taken from across all CPS Areas.
Please find an explanation of the terms used in the key findings at the bottom of this document.
Summary of findings
The updated analysis shows statistically significant disproportionality in our charging decisions during both time periods.
The increase in suspects in the updated analysis for the 2018-2022 time period has resulted in smaller odds ratios of likelihood of charge for all ethnic groups than reported by the original analysis.
In the most recent analysis looking at the 2022-2024 data, Non-British White, Mixed, Other and Black suspects, are more likely to be charged than White British suspects, and South Asian suspects are less likely to be charged than White British suspects.
Key findings
2018-2022 data
Table 1: unadjusted charge rates by grouped ethnicities
| Ethnicity | Charge rate (%) |
|---|---|
| Black | 80.4% |
| Mixed | 81.1% |
| Non-British White | 79.1% |
| Not Provided | 55.6% |
| Not Stated | 89.4% |
| Other | 77.3% |
| South Asian | 76.0% |
| White British | 76.1% |
Table 2: odds ratios for grouped ethnicities
| Ethnicity | Odds Ratios |
|---|---|
| Black | 1.15 |
| Mixed | 1.25 |
| Non-British White | 1.11 |
| Not Provided | 0.39 |
| Not Stated | 2.74 |
| Other | 0.98 |
| South Asian | 0.91 |
2022-2024 data
Table 3: unadjusted charge rates by grouped ethnicities
| Ethnicity | Charge rate (%) |
|---|---|
| Black | 82.9% |
| Mixed | 81.1% |
| Non-British White | 82.4% |
| Not Provided | 74.0% |
| Not Stated | 82.2% |
| Other | 82.0% |
| South Asian | 78.5% |
| White British | 78.6% |
Table 4: odds ratios for grouped ethnicities
| Ethnicity | Odds Ratios |
|---|---|
| Black | 1.24 |
| Mixed | 1.16 |
| Non-British White | 1.16 |
| Not Provided | 0.75 |
| Not Stated | 1.29 |
| Other | 1.13 |
| South Asian | 0.90 |
Explanation of terms
Odds ratios: Findings from the methodology used for these analyses are usually expressed as odds ratios. Odds ratios show us how likely suspects from different ethnic groups are to be charged compared to White British suspects. There are no White British odds ratios in tables 2 and 4 as this is the reference group that the other ethnic categories are being compared to.
An odds ratio of one means both groups are equally likely to be charged, greater than one means suspects in that group are more likely to be charged than White British suspects, and smaller than one means suspects in that group are less likely to be charged than White British suspects.
In general, odds ratios are preferred for this form of analysis because they provide more robust comparisons of the likelihood of events occurring relative to each other – they are explicitly designed for comparing outcomes such as the charge rates between White British and other ethnic groups.
Regression analysis: Regression analysis looks at the relationship between two or more variables and can determine the direction and strength of this relationship. An independent variable is the factor, predictor or driver that has an impact on the dependent variable or outcome. Regression allows researchers to control for the effects of one or more independent variables (e.g. sex or gender) while investigating the effect of another independent variable (e.g. ethnicity) on the dependent variable (e.g. charging).
Statistical significance: Statistical significance refers to the likelihood that a result can be attributed to a specific cause and did not just occur due to chance. A statistically significant difference implies that not only are the results unlikely to be the result of chance, but also that there is an effect or relationship between the variables being studied.
Unadjusted charge rates: Unadjusted charge rates are straightforward probabilities and do not adjust for any other factors such as crime type, gender, or age. They should be considered as the ‘raw’ likelihood of a charge.
Background and aims
Following the publication of the Disproportionality Action Plan in November 2024, the CPS committed to conduct further research to understand how to identify and address racial bias. We commissioned BIT (Behavioural Insights Team) to carry out two literature reviews to assess the evidence on best practice and how organisations in the criminal Justice system such as the CPS can reduce racial bias in decision making as well as enabling positive organisational culture change.
What works to reduce racial bias in criminal justice organisations?
The first report highlighted a number of factors evidenced in the literature which can help to reduce racial bias. These include:
- Increasing the diversity of staff within criminal justice organisations and strengthen awareness of systemic inequities
- Developing effective learning and training
- Supporting non-biased decision-making through protocols and guidelines
- Introducing technology-driven tools to reduce biased decision-making while maintaining awareness of their limitations.
- Considering incentives for engaging with anti-racism initiatives
- Creating strong accountability mechanisms
- Tracking progress and reviewing best practices regularly.
Enabling positive organisational culture in criminal justice organisations
The second report focused on organisational culture, meaning the beliefs, values, behavioural norms and practices that are identified with an organisation. It identified barriers and facilitators to changing organisational culture.
Some Barriers include:
- A lack of clear goals or an action plan
- A lack of a dedicated team responsible for change
- Closed communication
- A failure to track progress.
Some facilitators include:
- Creating an environment which facilitates change
- Setting clear expectations about a few important behaviours
- Reinforcing positive behaviours and taking action against instances of undesirable behaviour
- Embedding anti-racism into daily interactions
- Allowing staff to shape organisational change
- Embedding accountability for anti-racism into organisational structures
- Support staff through organisational culture change
- Tracking progress and reviewing best practices regularly
The full literature reviews are available to request from [email protected].
How the Crown Prosecution Service is tackling racial bias in the Criminal Justice System - our Action Plan