Research and analysis

A Profile of Repeat Offending by Children and Young People in England and Wales

Published 20 March 2025

1. Executive summary 

1.1 Introduction 

This report presents the findings of an exploratory analysis of repeat offending by children and young people in England and Wales. It focuses on one cohort of young people with repeated contact with the criminal courts, examining their characteristics, offending patterns, and needs. 

This analysis is based on newly linked administrative data made available via the Ministry of Justice (MoJ) Data First programme, funded by Administrative Data Research UK (ADR UK). Previous statistics have been limited to either reporting reoffending, or to exploring the needs of children in the criminal justice system. Leveraging advances in data linking, these statistics, for the first time have explored their intersection, providing critical new insights into the needs of those young people who are most frequently convicted. By providing a clearer picture of this group, the findings can help inform policies aimed at reducing reoffending and improving outcomes for children and young people in the justice system. 

1.2 Methodology 

The Data First criminal courts and offender assessment linked datasets formed the basis for this analysis. As MoJ administrative data does not contain unique personal identifiers, probabilistic matching was used to link individuals within and across these datasets. 

Data sources 

To construct a complete court history for each young person, magistrates’ court data (2011-2023) and Crown Court data (2013-2023) were brought together. Additionally, for the first time, records from the Offender Assessment System (OASys) were also linked to enable analysis of a broad range of characteristics and needs, including to identify key differences between prolific and nonprolific young offenders. 

Defining prolific young offenders 

This analysis focuses on a cohort of children and young people who have been involved in repeat offending. For the purposes of this analysis, these are defined as individuals who: 

1. were 18 years old in 2019 and had a conviction for an offence committed that year (the index conviction); and, 

2. had been convicted in a criminal court at least three times for offences committed between the ages of 10 and 17 (youth convictions). 

For clarity and consistency, this group is hereafter referred to as ‘prolific young offenders’ throughout the report. 

The definition used in this report is not the same as the those used in other published Official Statistics[1] on prolific offenders.  These include cautions from the Police National Computer (PNC) that are not recorded in criminal courts datasets. 

1.3 Findings 

Risks and needs

• For offenders assessed at age 18, prolific young offenders had higher levels of need compared to nonprolific young offenders across all eight criminogenic areas: accommodation, employability, relationships, lifestyle, drug misuse, alcohol misuse, thinking & behaviour, and attitudes.
• Each area of criminogenic need is comprised of a set of questions. The questions which exhibited large differences in significant need include:
  • school attendance (49% prevalence in prolific young offenders, compared to 20% of nonprolific young offenders);
  • susceptibility to influence by criminal associates (48% prevalence in prolific young offenders, compared to 21% of nonprolific);
  • a history of childhood behavioural problems (52% prevalence in prolific young offenders, compared to 24% of nonprolific young offenders)
• A larger proportion (27%) of prolific young offenders live in the most deprived 10% of neighbourhoods. This compares to 21% of nonprolific young offenders and 11% of all 18-year-olds across England.

Nature and frequency of offending

• Most prolific young offenders (70%) were first convicted for offences committed between ages 14 and 16; 22% were convicted for an offence committed before age 14.
• Whilst all young people in the cohort met the threshold of three youth convictions before turning 18 years old, 28% had exactly three, 43% had four to six, 18% had seven to nine, and 11% had ten or more youth convictions. This suggests that even among prolific young offenders, there is a subset that is particularly criminally active.
• Over a quarter (27%) of prolific young offenders received at least one immediate custodial sentence before the age of 18 years old.
• For prolific young offenders, theft offences accounted for 23% of youth convictions, falling to 12% of offences committed as adults. Conversely, convictions for drug and violence offences increased from 16% for youths to 25% for adults.
• The more youth convictions an offender had, the more likely they were to continue offending frequently as adults. 9% of prolific young offenders were convicted of 8 or more offences as young adults between the ages of 18-22.

1.4 Limitations 

While this analysis provides valuable new insights, there are limitations that should be considered when interpreting findings. 

Data coverage: 

• Magistrates’ court data spans the period from January 2011 to March 2023, while Crown Court data is available from January 2013 to March 2023. As a result, magistrates’ court cases committed to the Crown Court before 2013 may lack outcome information, potentially leading to a slight undercount of prior offending and possible exclusion of some individuals from the prolific offender cohort. However, this impact is minimal, as it would only affect cases where a child aged 10-12 was referred to the Crown Court. 

• As part of reforms in His Majesty’s Courts and Tribunals Service (HMCTS), new administrative data systems have been introduced. While some cases remain recorded in legacy systems, cases recorded in the new Common Platform system (from mid-2021 onwards) are not included in this analysis. This likely leads to the exclusion of more recent cases; however, by establishing a baseline in 2019, the impact of this system change is limited to adult offences within this analysis. 

Record linkage: 

• As individuals do not have unique identifiers within or across datasets, the data-linking software Splink[2] was used to probabilistically match records. While this approach enables linking within and between datasets, there is a small risk of incorrect matches or missed links. 

• Splink was also used to link offender assessment data where available.  Not all offenders are assessed however, and the likelihood of an offender having an assessment also depends on the type of offence and sentence received. 

Further details on the data linkage process are provided in the accompanying Technical Guide. 

2. Introduction 

2.1 Data First 

Data First[3] is a pioneering data-linking, research and academic engagement programme led by the Ministry of Justice (MoJ) and funded by Administrative Data Research UK (ADR UK)[4]. By linking administrative datasets from across the justice system, Data First is maximising the use and impact of data that is already collected when people interact with justice services. 

Since the inception of the programme in 2019, MoJ has linked eight justice datasets that connect cases and people from across the criminal, family and civil courts, prisons, probation services and offender assessment. This creates opportunities for analysis that have not been possible before, building a more holistic picture of justice system users, their journeys, and outcomes as they interact across different jurisdictions of the system. The data enables researchers to identify patterns – such as frequent system engagement or common transitions between services – to better understand the impact of justice services and interventions, and also ‘what works’ to improve justice outcomes. 

These deidentified datasets are made available to accredited researchers across government and academia. By working in partnership with academic experts, Data First is delivering vital new evidence for policy and practice insights to deliver departmental priorities. 

2.2 Background 

Prolific offenders have a disproportionate impact on the justice system. A recent MoJ report (2023)[5] found that while they make up just 9% of the offending population, they accounted for over half (52%) of all convictions between 2001 and 2021. Given their frequent interactions with the justice system, further understanding of this group can provide valuable insights into their offending patterns and associated needs. 

This report makes use of newly linked data from the Data First programme.  For the first time this connects criminal court records with offender assessment data. This linkage offers a unique opportunity to examine young people with repeat convictions – shedding light on their demographics, offending journeys, and identified needs in ways that were not previously possible. 

2.3 Official Statistics 

The MoJ publishes Official Statistics on both prolific and young offenders. The Characteristics of Prolific Offenders[6] report provides the latest statistics on prolific offending patterns, drawing on data from the Police National Computer (PNC). 

The Proven Reoffending Statistics[7] collection produces quarterly figures on reoffending patterns, including data on young people (referred to there as juvenile offenders). Additionally, the Youth Justice Board publishes annual Youth Justice Statistics[8], which provide Accredited Official Statistics on children in the justice system in England and Wales. 

The Youth Justice Board has also assessed the needs of sentenced children in the Youth Justice System[9]. These experimental statistics explored a subset of data obtained from AssetPlus, an assessment and planning framework for use with children by Youth Offending Teams. 

The findings in this report are not directly comparable to other published statistics due to different units of data, processing, and analysis. Unlike the Characteristics of Prolific Offenders report, this analysis does not use PNC data and therefore excludes cautions, relying solely on court convictions to measure prolific offending. A new feature of this analysis is the use of offender assessment data, which provides insights into the needs of offenders – something not covered in existing relevant statistics. 

3. Methodology 

3.1 Data sources 

This analysis was conducted using the Data First magistrates’ court, Crown Court, offender assessment, and cross-justice linking datasets. These datasets provide detailed information on criminal court cases and offender needs in England and Wales. 

• Dataset 1 - Ministry of Justice Data First magistrates’ courts defendant - England and Wales: Provides data on people within the magistrates’ courts system in England and Wales and has been constructed using weekly extracts from LIBRA originally used for Home Office Court Appearance Statistics (HOCAS). The dataset covers all court appearances from 1 January 2011 to 31 March 2023.      

• Dataset 2 - Ministry of Justice Data First Crown Court defendant - England and Wales: Provides data on defendants’ appearances in criminal cases before the Crown Court in England & Wales from 2013, and has been extracted from XHIBIT management information system, used by His Majesty’s Courts and Tribunals Service (HMCTS) to manage cases within the Crown Court.  The dataset covers all court appearances from 1 January 2013 to 31 March 2023. 

• Dataset 3 – MoJ Data First Offender Assessment dataset - England & Wales: Provides data on offender assessments recorded for service users in custody and in the community system in England and Wales from 2011. The data has been extracted from the Offender Assessment System (OASys), used by His Majesty’s Prison & Probation Service (HMPPS) in England and Wales to measure the risks and needs of offenders in custody or under supervision in the community.  The dataset covers all offender assessments from 1 January 2011 to 31 December 2023 

• Dataset 4 - Ministry of Justice Data First cross-justice system linking dataset - England and Wales: Enables records that refer to the same people in separate justice datasets to be joined.  The dataset contains records from 1 January 2011 to 31 March 2023.   

For further details on datasets used, see the accompanying technical guide. 

3.2 Data linking 

The datasets used for this analysis do not contain a consistent unique identifier to enable linkage between them. To address this, records were linked at person level using Splink, a probabilistic matching tool developed by MoJ for deduplicating and linking large datasets. This makes it possible to track individuals within and across datasets for the first time, allowing for the identification of repeat users and providing a clearer picture of their interactions across the justice system. 

More details on the data linking process and use of Splink can be found in the Technical Guide. 

3.3 Defining prolific young offenders 

This analysis focuses on a cohort of children and young people who have been involved in repeat offending. For the purposes of this analysis, these are defined as individuals who: 

1. Were 18 years old in 2019 and had a conviction for an offence committed that year (the index conviction). 

2. Had been convicted in a criminal court at least three times for offences committed between the ages of 10 and 17 (the youth convictions). 

In contrast, nonprolific young offenders were defined as individuals who: 

1. Were 18 years old in 2019 and had a conviction for an offence committed that year (the index conviction). 

2. Had been convicted in a criminal court no more than twice for offences committed between the ages of 10 and 17 (the youth convictions). 

For clarity and consistency, these groups are hereafter referred to as either ‘prolific’ or ‘nonprolific young offenders’ throughout the report. 

These definitions were developed using a data-driven approach: 

• Focusing on individuals who were 18 in 2019 allowed for the inclusion of complete criminal court histories (ages 10-17), as the Data First magistrates’ court dataset covers the period from 2011. It also allowed a reasonable period to capture subsequent offending following the index conviction given coverage through to 2023. 

• Requiring an adult conviction at age 18 ensured that offender assessments could be included, as these are only conducted on adults. 

The definition used in this report is not the same as the those used within other published Official Statistics[10] on prolific offenders, as these include cautions from the Police National Computer (PNC) that are not recorded in criminal courts datasets. 

Using this methodology, 10,027 young offenders were identified and analysed, split into prolific and nonprolific groups (see Table 1). The findings from this analysis are presented in Section 4. 

Table 1: Size of young offender cohorts 

Cohort	Number of Offenders
Prolific Young Offenders	1,266
Non-prolific Young Offenders	8,761

3.4 Offender assessments 

To better understand the characteristics of prolific young offenders, this analysis uses data from the Offender Assessment System (OASys), a tool used by His Majesty’s Prison & Probation Service (HMPPS) to assess offender risks and needs and support them during custody and in the community. 

Through OASys, assessors have access to a range of different tools. This analysis uses information from the core assessment templates, of which two are in active use: 

• Layer 1 (basic) assessments are a significantly shortened version which do not assess an offender’s needs and are limited to a basic sentence plan. 

• Layer 3 assessments (hereafter referred to as ‘full assessments’) contain the full range of questions in relation to the risks and needs related to an individual’s offending behaviour. 

Only full assessments were included in this analysis as they contain information on children and young people’s needs. 

Full assessments were available for 65% of prolific young offenders and 17% of nonprolific young offenders at age 18 or 19. The lower proportion for nonprolific offenders reflects that assessments are only conducted for those receiving custodial or community sentences, meaning that many nonprolific offenders were not assessed. 

Further details on offender assessment data linkage and methodology can be found in the Technical Guide. 

4. Findings 

4.1 Characteristics 

This section examines the demographic and socioeconomic background of prolific young offenders, focusing on sex, ethnicity, and levels of deprivation. Understanding these characteristics helps provide context for their experiences within the justice system and may highlight patterns linked to prolific offending. 

Sex 

Young men made up the vast majority of prolific young offenders, accounting for 94% of the group. This was noticeably higher than in the non-prolific group, where 79% were male[11]. This suggests that young men were not only more likely to offend, but more likely to become repeat youth offenders compared to their female counterparts. 

Ethnicity[12] 

Compared to the national population of 18-year-olds in England and Wales (2021)[13], young people from Black and Mixed ethnic backgrounds were over-represented in the prolific young offender cohort, with Black individuals being the most over-represented. In contrast, young people from Asian backgrounds were under-represented. 

 

A comparison of ethnicity group breakdown with non-prolific young offenders was explored but not included in this report due to the high level of missing ethnicity data amongst this group (45%). 

Deprivation[14] 

Figure 3 shows the proportional distribution of prolific young offenders, non-prolific young offenders, and the overall 18-year-old population across deprivation deciles in England. Areas of residence are assigned to deciles based on levels of relative deprivation. 

Prolific young offenders were more likely than both non-prolific young offenders and the general 18-year-old population to live in deprived neighbourhoods at age 18, indicating greater socioeconomic disadvantage amongst the group. Over a quarter (27%) of prolific young offenders lived in the 10% most deprived areas, compared to just over one in five (21%) non-prolific young offenders and one in nine (11%) 18-year-olds. 

4.2 Offending patterns 

This section looks at how offending patterns develop over time for prolific young offenders. It starts with their first offences as children and then examines how their offending changes as they get older. It also compares how prolific and non-prolific offenders are sentenced. 

4.2.1 Early contact with the justice system 

Age at first conviction 

Figure 4 shows the age at which prolific young offenders committed the offence that led to their first conviction. 

 

Some of these children began offending at a young age, with more than one in five (22%) committing their first offence before turning 14. Most received their first conviction for an offence committed between the ages of 14 and 16, which accounted for 70% of the prolific offenders group.  At age 17, a further 9% committed their first offence, meaning that nearly one in ten in the cohort went on to meet the prolific offender threshold (three or more convictions) within just one year.  

Frequency of youth offending 

While all young people in the prolific young offender cohort met the threshold of at least three youth convictions, Figure 5 highlights substantial variation in rates of youth offending. 

Nearly one in three (29%) prolific young offenders had more than twice the number of convictions required to meet the ‘prolific’ definition, with seven or more convictions before turning 18 (compared to the threshold of three). Around a third of this group (11% of the total) had ten or more youth convictions, suggesting that even among prolific young offenders, there is a subset that is particularly criminally active. 

Sentencing for early offences 

Figure 6 shows the types of sentences given to prolific young offenders for their first three convictions. Community sentences were the most common outcome, given in 1,072 (85%) first convictions and 880 (70%) third convictions, declining as courts made use of further sentencing options. As the use of community sentences declined for subsequent offences, the use of fines (2% of first sentences compared to 5% of third sentences) and those otherwise dealt with[15] (8% to 16%) both increased. The proportion of offenders given immediate custody increased from 2% for the first conviction to 4% by the third. 

 

The figures in Table 2 show the number of offenders who received each type of sentence across any of their first 3 convictions, and across sentences for all of their youth convictions. The figures differ from those presented in Figure 6, as each offender is counted only once within each sentence category. Of the 1,266 prolific young offenders, 344 (27%) received at least one immediate custodial sentence for an offence committed before they turned 18. Almost all (99%) prolific young offenders received at least one community sentence for any of their youth convictions. 

Table 2: Number and proportion of prolific young offenders receiving each type of sentence at least once [t2] 

Sentence type	First 3 convictions: Number of offenders	First 3 convictions: Percentage of cohort	All youth convictions: Number of offenders	All youth convictions: Percentage of cohort
Absolute/conditional discharge	160	13%	331	26%
Community sentences	1,245	98%	1,250	99%
Fine	102	8%	238	19%
Immediate custody	99	8%	344	27%
Otherwise dealt with	390	31%	675	53%

[t2]Percentage columns are calculated against the 1,266 prolific young offenders in the cohort. As each prolific young offender can receive different sentence types, percentages will not sum to 100%. 

While figure 6 showed that community orders were the most common type of sentence handed to prolific young offenders, Figure 7 provides a further breakdown of the types of community orders issued by the magistrates’ court for the prolific offender’s first three youth convictions. 

Referral orders[16] were the most common sentence for 10 to 17-year-olds, particularly for first-time offenders. For their first conviction, 75% of offenders received a referral order, which requires them to complete a rehabilitative and restorative contract as part of their sentence. 

As the number of previous offences increase, a wider range of sentences are used by the courts. Youth rehabilitation orders, which allow courts to directly impose specific requirements, became increasingly common. By the third conviction, 477 youth rehabilitation orders – including those with intensive supervision and surveillance – were issued.  This accounts for over half (56%) of all community sentences issued by the magistrates’ court and were the most frequently used sentence at this stage (38% of all sentences). 

4.2.2 Transition to adulthood 

This section compares the types of offences committed as children to those committed as young adults, while exploring patterns between the frequency of offending as youths and adults. 

Changes in offence types from youth to adulthood 

The types of offences committed by prolific young offenders changed as they moved from youth to adulthood. Figure 8 shows that summary non-motoring offences were the most common type of offence in both youth and adulthood, accounting for a third (33%) of youth convictions and 32% of adult convictions. Summary offences are generally less serious offences, dealt with at the magistrates’ court. 

However, there were changes in other offence types. Theft was the second most common offence for youths (23%) but dropped to the fourth most common offence amongst adults (12%). Drug-related and violence against the person offences, however, became more common, rising from 16% amongst youths to 25% in adulthood. 

Another noticeable change was the increase in motoring offences, as young people reached driving age. Consequently, the decline in the proportion of convictions in adulthood for some indictable and triable-either-way offences – offences which are more likely to be heard at the Crown Court and where more severe sentences can be imposed - may partly reflect this rise in motoring-related offences. 

Link between youth and adult offending 

When examining youth offending, Figure 5 categorised prolific young offenders based on their number of youth convictions. Figure 9 builds on this by showing how the proportion of youth offenders who went on to be convicted of higher numbers of adult offences (five to seven, and eight or more) increases incrementally in line with increasing numbers of youth convictions. 

Whilst 6% of prolific young offenders who had three youth convictions went on to have eight or more adult convictions, this increased to 8% of those with four to six youth convictions. For offenders with seven to nine youth convictions 12% were convicted 8 times as adults, and for those with ten or more youth convictions, this increased to 16%.  A similar increasing pattern can be observed in prolific young offenders who went on to have the second highest count of five to seven adult convictions. 

Conversely, whilst 20% of prolific young offenders with three youth convictions went on to have one adult conviction, this reduced to 19% of those with four to six youth convictions, reducing further to 15% of those with seven to nine, and to 10% of those with ten or more youth convictions.  Again, a similar reducing pattern can also be observed in prolific young offenders who went on to have the second lowest count of two to four adult convictions. 

This suggests the more youth convictions an offender had, the more likely they were to continue offending prolifically into adulthood.

4.2.3 Adult offending patterns 

This section presents the frequency of convictions and types of offences committed as young adults, exploring how sentence outcomes change into adulthood. Comparisons to non-prolific offenders are also explored showing how offending and sentencing for prolific young offenders as young adults differ to non-prolific offenders. 

Frequency of adult offending 

As highlighted in Section 3.3, all offenders included in this analysis had at least one adult conviction – the index conviction. However, most people in the cohort continued to offend after that. 

Figure 10 shows that 82% of the cohort received more than one conviction for offences committed between the ages of 18 and 22. Just over a quarter (27%) had five of more adult convictions, with a third of these, or almost one in ten (9%) overall receiving eight or more. 

Sentencing for youths and adults 

Sentencing patterns also changed once offenders reached adulthood. As shown in Figure 11, fines (38%) and immediate custody (23%) became the most common outcome for offences committed between the ages of 18 and 22. Unlike in youth sentencing, which has a distinct and separate sentencing framework to adult sentencing[17], community sentences were no longer the most common outcome. 

 

Comparing prolific and non-prolific offenders in adulthood 

 

Differences in sentencing 

Figure 12 compares the sentences given to prolific and non-prolific young offenders for their first adult conviction. 

Previous findings of guilt are an aggravating factor in sentencing[18], and this is reflected when comparing sentences of prolific and non-prolific young offenders. 20% of prolific offenders received immediate custody for their first adult conviction, compared to 5% of non-prolific offenders. For non-prolific offenders, fines were more common, with 60% receiving this sentence type compared to 24% of the prolific group. Community sentences were given to 30% of prolific offenders for their first adult conviction, while only 18% of non-prolific received this type of sentence. 

Differences in offence types 

These differences in sentencing are also partly explained by the types of offences committed. Figure 13 shows that non-prolific offenders were more likely to commit less serious offences, with nearly two-thirds of their first adult convictions being summary offences (34% summary motoring, 33% summary non-motoring). 

Prolific offenders, however, were more likely to be involved in indictable/triable either-way offences, such as drug offences and theft. More than a third (35%) of first adult convictions for prolific offenders were for theft and drug offences. The only indictable triable-either-way offence category less common in the prolific group was sexual offences. 

4.3 Offender needs 

This section explores the needs of prolific young offenders identified from their full offender assessments, using data extracted the Offender Assessment System (OASys). 

A full assessment contains a large range of sections including offence details, criminogenic needs, responsivity needs, risk of serious harm assessment, risk management plan and sentence plan. For this analysis, we are interested in criminogenic needs, which are needs which can contribute to a person reoffending. There are eight criminogenic needs sections measuring factors linked to offending behaviour; Accommodation, Employability, Relationships, Lifestyle, Drug misuse, Alcohol misuse, Thinking & Behaviour, and Attitudes. Each section is made up of a range of questions related to the section’s theme. 

Overall rates of need 

Figure 14 summarises questions from each section to show the overall rates of identified criminogenic need for those prolific and non-prolific young offenders with a full offender assessment. Details for how overall rates of need were calculated can be found in the accompanying technical guide. 

Among those with an assessment, prolific young offenders had higher levels of need across all eight assessment sections. The most substantial difference was in attitudes. This explores pro-criminal attitudes, attitude to supervision/license, attitude to community/society, and motivation to address offending behaviour. Of the prolific cohort, 84% had an identified need relating to attitudes, compared to 58% of the non-prolific cohort. 

The most prevalent area of need for all young offenders was ‘lifestyle and associates’. This includes risk-taking behaviour, association with criminal peers, and engagement in activities that encourage offending. Among prolific offenders with an assessment, 88% had an identified need in this category, compared with 71% of non-prolific offenders. 

Data-driven insights 

To identify the areas where prolific young offenders had the most pronounced differences in need compared to their non-prolific peers, a data-driven approach was adopted. Each of the eight criminogenic need sections and three responsivity need sections are comprised of a set of questions, answers to which are recorded in many different formats. To compare them, we calculate standardised mean differences (SMDs)[19], which allow us to identify the questions where need differs the most between the two cohorts. 

SMDs measure how much two groups differ, while accounting for differences in variability between questions, and places these on a comparable scale. Originally devised by David Cohen (and commonly referred to as Cohen’s D effect sizes), he described values between 0.2 and 0.5 to be “small effects”, 0.5-0.8 as “moderate effects” and values above 0.8 as “large effects”. 

Table 3: Offender assessment questions where prolific young offenders had the greatest increase in need compared to non-prolific young offenders 

	Prolific	Non-prolific	SMD
Number in group	816	1,512
Need
Offender has significant problems with any regular activities that encourage offending	47%	18%	0.68
Offender has significant issues with school attendance	49%	20%	0.67
Offender has significant problems of being easily influenced by criminal associates	48%	21%	0.63
Offender has evidence of childhood behavioural problems	52%	24%	0.61
Offender has significant problems with pro-criminal attitudes	34%	11%	0.61
Offender has problems with recklessness and risk-taking behaviour	59%	33%	0.56
Offender experienced significant problems during childhood	49%	25%	0.54
Offender has financial issues linking to their offending behaviour	67%	41%	0.52
Offender has significant historical issues with employment	44%	21%	0.52

In this analysis there were no “large effects” (or scores above 0.8). However, “moderate effect sizes”, or SMDs above 0.5 were found, and this threshold is used to flag areas where prolific offenders had notably higher needs. This focuses on the largest standardised differences rather than absolute percentage differences. Questions where significant needs were high for prolific offenders and SMDs were greater than 0.5 are presented in Table 3[20], with a full table of SMD values for all questions available in the accompanying standardised mean differences table. 

Questions where significant need has comparatively risen the most include: 

• problems with regular activities that encourage offending, with 47% prevalence in the prolific offending group compared to 18% of non-prolific offenders and a standardised difference of 0.68; 

• school attendance, with 49% of prolific young offenders presenting significant problems in this area, compared to 20% of non-prolific young offenders (a standardised difference of 0.67); 

• susceptibility to influence by criminal associates, with 48% prevalence in prolific young offenders, compared to 21% of non-prolific, and a standardised difference of 0.63; and 

• a history of childhood behavioural problems, with 52% prevalence in prolific young offenders, compared to 24% of non-prolific young offenders. 

Prolific young offenders were also more likely to have a history of pro-criminal attitudes, a tendency toward recklessness, problems during childhood, previous difficulties with employment and financial issues linked to their offending behaviour. 

Figures 15 and 16[21] capture the full breakdown of recorded needs for these nine questions, showing that prolific young offenders had substantially higher rates of need across all questions. 

Assessments of specific key needs   

While data-driven analysis identifies the largest standardised differences, additional questions were also explored to provide further insight into key needs and characteristics of prolific young offenders. 

Analysis was carried out on ten full assessment questions allowing for direct comparisons between the two cohorts. Rather than adopting the data-driven approach above, which used standardised mean differences (SMDs) to compare levels of significant need between the two groups, Table 4 presents the rates of need for prolific and non-prolific young offenders more broadly across these ten questions, aggregating both ‘some’ and ‘significant’ need, where relevant.  As these questions have been chosen due to their general relevance, SMDs have not been calculated or presented. 

Table 4: Comparison of key OASys needs for prolific and non-prolific young offenders. 

	Prolific	Non-prolific
Number in group	816	1,512
Need
Has some or significant issues with permanence of accommodation	50%	31%
Has some or significant emotional well-being coping difficulties.	64%	51%
Has been on medication for mental health problems in the past.	10%	9%
Has any self-harm, attempted suicide, suicidal thoughts or feelings	31%	28%
Has some or significant issues with school attendance.	80%	51%
Has some or significant learning difficulties.	28%	19%
Has some or significant current problems with relationships or close family members	70%	49%
Evidence that offender is involved in current or previous domestic violence/partner abuse.	27%	23%
Has some or significant problems with their financial situation.	67%	49%
Was influenced by peer groups.	49%	35%

In all areas, prolific young offenders exhibited higher levels of need. They were more likely to have issues with permanence of accommodation and 70% had some or significant identified problems with relationships or close family members. 80% of young prolific offenders had recorded issues with school attendance compared to 51% of non-prolific offenders. 

5. Limitations and caveats 

5.1 Data source limitations 

Temporal coverage 

The Data First magistrates’ court dataset includes cases from 1 January 2011 to 31 March 2023, meaning that full court histories are available for the young offender cohorts used in this analysis. However, the Data First Crown Court dataset covers a shorter period, from 1 January 2013 to 31 March 2023. 

As a result, magistrates’ court cases committed to the Crown Court before 2013 may lack outcome information, potentially leading to a slight undercount of prior offending and possible exclusion of some individuals from the prolific offender cohort. However, this impact is minimal, only affecting cases where a child aged 10-12 was referred to the Crown Court.  This could also result in a magistrates’ court case being linked to an incorrect Crown Court case. 

System coverage 

The magistrates’ court and Crown Court datasets originate from the LIBRA/HOCAS and XHIBIT legacy case management systems used by His Majesty’s Courts and Tribunal Service (HMCTS). As part of HMCTS reforms, new administrative data systems have been introduced. From mid-2021, most new criminal cases have been recorded in the new case management system Common Platform.  These cases are not yet included in Data First datasets or in this analysis. 

As a result, dataset coverage has decreased over time, particularly for cases received after 2021. While most cases disposed of in 2021 and 2022 are still captured, some court appearances – particularly adult offences – may be missing or underestimated. By establishing a baseline in 2019, the impact of this system change has been mitigated. 

Administrative data 

As this analysis relies on administrative databases, it is subject to inaccuracies common in large-scale data recording systems, such as typographical errors and missing data. Missing data is labelled as “Unknown” in this analysis, and results should be interpreted with caution. 

This data was extracted from systems designed for operational administration rather than research. As a result, clerical and input errors may affect data quality, record linkage and deduplication. 

5.2 Methodological limitations 

Person-level linkage 

Administrative justice datasets do not contain unique identifiers within or across datasets. Instead, probabilistic matching using the software Splink was used to estimate person-level linkages. While this method is designed to accurately link administrative records, challenges inherent to the data remain. These include: 

• Typographic or phonetic errors in names or other identifying fields 

• Changes in names or addresses over time 

• Use of aliases, nicknames, or diminutives 

• Missing or incomplete data 

Given these challenges, some false links may exist – where records belonging to multiple individuals are incorrectly assigned to a single person. Similarly, some true links may have been missed due to matching probability thresholds or data quality issues. 

Linking magistrates’ court and Crown Court cases 

Probabilistic matching was used to link magistrates’ court cases to Crown Court cases. For situations where a magistrates’ court case mapped to multiple Crown Court cases, a further one-to-one case linkage methodology was developed. If multiple Crown Court cases appeared to be the best match, the case with a conviction was prioritised. If no clear best match could be identified, then no Crown Court record was retained. 

Of the 46,653 magistrates’ court cases analysed, 7,426 were committed to Crown Court, either for trial or sentencing. Of these: 

• 6,599 cases (88%) had at least one probabilistically linked Crown Court case, of which, 

• 6,414 (86%) had an adequate best match using the linkage methodology. 

Offender assessment linkage 

To assess offender needs, this analysis used linked data from the Data First offender assessment dataset. Since offenders can receive multiple offender assessments over time, a methodology was developed to select the most relevant assessment – the details of which can be found in the Technical Guide. 

This approach means that some individuals did not have a full assessment included in the analysis – either because no assessment was linked or because it was initiated too long after age 18 to reliably reflect the young person’s needs at that time. 65% of prolific young offenders had a linked assessment, compared with only 17% of non-prolific young offenders.   

Offender assessments are primarily conducted on individuals serving sentences in custody or in the community. Non-prolific young offenders were less likely to receive an assessment because they were more likely convicted of less serious offences which would not typically result in an assessment. 

As a result, the 17% of non-prolific offenders with a full assessment are likely skewed towards more serious offences. This means that comparisons of offender needs between prolific and non-prolific young offenders may underestimate the true differences between the two groups. 

Longitudinal comparisons 

The findings in this report are based on a single cohort of prolific offenders who were 18 years old in 2019. The findings from this analysis may not reflect groups of young offenders at other points in time. 

Errors in statistical systems and processes 

Although every effort is made to ensure accuracy, errors can occur in statistical processes. To minimize this risk, all procedures have been reviewed and thoroughly quality assured. 

5.3 Comparisons to other published statistics 

The findings in this report are not directly comparable to other published statistics due to different units of data, processing, and analysis. The Characteristics of Prolific Offenders report defines offenders using Police National Computer (PNC) data, which includes both convictions and cautions. In contrast, this analysis does not use PNC data and relies on criminal court records to identify prolific offenders. This means that cautions, which might have contributed to prolific offending patterns in the Characteristics of Prolific Offenders, are not included. This difference should be considered when interpreting and comparing findings. 

6. Contacts 

Media contacts 

Press enquiries should be directed to the Ministry of Justice press office. 

020 3334 3536 

Statistical contacts 

Other enquiries about these statistics should be directed to Data First at the Ministry of Justice: datafirst@justice.gov.uk 

Ministry of Justice 

Data Directorate: Data First: Research, Academic Engagement, Communications 

10th Floor 

102 Petty France 

London 

SW1H 9AJ

References

[1] Characteristics of Prolific Offenders, 2000-2021 

[2] Splink is the data linking software development by the Ministry of Justice.

[3] https://www.gov.uk/guidance/ministry-of-justice-data-first 

[4] https://www.adruk.org/our-work/browse-all-projects/data-first-harnessing-the-potential-of-linked-administrative-data-for-the-justice-system-169/ 

[5] Ministry of Justice, released February 2023, GOV.UK, Characteristics of Prolific Offenders, 2000-2021, page 3 

[6] Characteristics of Prolific Offenders, 2000-2021 

[7] Proven reoffending statistics - GOV.UK 

[8] Youth Justice Statistics - GOV.UK 

[9] Youth Justice Board/Ministry of Justice, Published 28 May 2020, GOV.UK, Assessing the needs of sentenced children in the Youth Justice System - GOV.UK

[10] Ministry of Justice, released February 2023, GOV.UK, Characteristics of Prolific Offenders, 2000-2021

[11] 5% of the non-prolific young offender cohort did not have a recorded sex 

[12] Self-reported ethnicity data was used, rather than police reported ethnicity. 

[13] Office for National Statistics (ONS), released 23 January 2023, dataset, Ethnic group by age and sex in England and Wales 

[14] Based on Indices of Multiple Deprivation data for England from 2019

[15] Otherwise dealt with includes disposals such as suspended sentences, one day in police cells, disqualification order, restraining orders, confiscation orders, travel restriction orders, disqualification from driving, hospital orders, guardianship orders, recommendation for deportation and other disposals. 

[16]  A referral order is an order available for young offenders who plead guilty to an offence and require that an offender must agree a contract of rehabilitative and restorative elements to be completed within the sentence Referral order – Sentencing

[17] Young people and sentencing – Sentencing 

[18] Aggravating and mitigating factors – Sentencing

[19] For this analysis, standardised mean differences were calculated as described in: 

Austin, P.C., 2009. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity‐score matched samples. Statistics in medicine, 28(25), pp.3083-3107. 

[20] Full assessment questions that were directly related to the cohort definition, such as ’Indicates if the current offence(s) are part of an established pattern of similar offending’, were excluded from this table. 

[21] In Figure 16, the first question (behavioural problems) is scored on a yes/unknown scale, meaning that unknown captures both ‘no need’ and ‘unknown need’. Contrastingly, the second question (financial issues) is scored on a yes/no/unknown scale.