Female reoffending trajectories in England from 2000 to 2020: A longitudinal administrative data study

Female reoffending trajectories in England from 2000 to 2020: A longitudinal administrative data study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Female reoffending trajectories in England from 2000 to 2020: A longitudinal administrative data study Christabel Coles, Alice Wickersham, Darrick Jolliffe, Nigel Blackwood, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8115965/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract As females are a minority group in the criminal justice system, they have often been neglected in criminological research. Research exploring offending trajectories has largely focused on male and mixed-gender samples and studies using female-only cohorts have been constrained by small sample sizes, short follow-ups and limited evidence from England. However, as rates of female imprisonment increase, it is ever more important that we attempt to understand females’ engagement with the criminal justice system. This study used a linkage of the Police National Computer and the National Pupil Database to create a birth cohort of female offenders in England (N=196,089). The study explored their offending trajectories from the age of criminal responsibility (10 years) up to age 31 years. A latent class analysis identified a four-class model with the following classes: ‘Life-course-persistent’ (11%), ‘Adolescent-limited’ (54%), ‘Young-adult-limited’ (16%) and ‘Adult-onset’ (19%). These classes resemble those identified using male and mixed-gender samples. By identifying the key developmental patterns of female offending in England, this study provides an empirical basis for early identification and tailored support for those females most at risk of persistent offending. Social science/Criminology Biological sciences/Psychology Social science/Psychology Introduction The UK has one of the highest female imprisonment rates in Western Europe 1 . Women who are imprisoned have a significantly elevated prevalence of mental disorder, substance misuse, homelessness and traumatic life experiences 2-6 . Imprisonment may serve as a re-traumatising experience 7 , with female prisoners’ rates of self-harm nearly five times higher than males’ 8 . In addition, in 2020, over 17,500 children were estimated to have been separated from their mother as a result of maternal imprisonment 9 . Such separations are associated with insecurities of attachment, depression and an increased risk of antisocial behaviour 10-12 . Despite the notable differences and impacts there has been little research about female offending and this is particularly true in England and Wales 13 . This may be because women tend to make up a small proportion of offenders, and their offences are on average less serious 14 , but the result of this absence is that most criminological theories and indeed criminal justice approaches are simply adapted from research and knowledge about males. We must therefore enhance our understanding of the developmental trajectories of female reoffending and their associated risk factors to inform prevention and intervention strategies, including reduced imprisonment and increased community-based sentencing strategies in this group 15 . Developmental trajectories of offending explore the antecedents and course of offending behaviours across the lifespan 16 . Moffitt drew on developmental trajectory work to suggest that two main offending trajectories exist: life-course-persistent (LCP) and adolescent-limited (AL) 17 . LCP offending is characterised by an early onset of serious and diverse offending that persists across the life course, associated with a combination of neurodevelopmental and environmental risk factors. AL offending is characterised by less serious offending, mainly during the adolescent years, associated with a delinquent peer group and a gap between biological and social maturity. While both groups of offenders were predominantly male, female offenders were more typically observed in the adolescent-limited group. Since Moffitt’s publication, life course trajectories of male offending from childhood into adulthood and their correlates have been well researched 18 . Methods used to identify groups have varied, with some studies using finite mixture modelling approaches 19 and others using hard-coding methods and algorithms 20 . Finite mixture modelling is a data-driven approach that estimates groups probabilistically, whereas hard-coding and algorithm methods use pre-defined rules or thresholds and do not account for population heterogeneity. Such work has cumulatively suggested that additional trajectories beyond LCP and AL offending exist, including late-onset escalating and early-onset desistant trajectories 21 . The exact number of groups identified appears to be moderated by methodological factors: for example, studies which used self-report rather than official data identified more offending trajectories 21 . To date, most research exploring offending trajectories has been performed using male or mixed-gender samples (see 21-23 for reviews). Recent work using a mixed-gender birth cohort from the linked Police National Computer and National Pupil Database identified five offending trajectories (including an LCP, two ALs, and two late-onset) 24 . However, these findings are likely to have been driven by the large proportion of male offenders in the sample. True heterogeneity in female re-offending patterns and characteristics of trajectories may be masked in such studies due to higher rates of male offending 25 . Overall, research on female trajectories has found similar trajectory types to those in males but has also identified some important differences. While research has found female LCP groups 26-28 , women began offending at a later age 29,30 , committed fewer violent and serious offences 31 and were less likely to follow persistent offending trajectories compared to men. Some studies also find adult-onset trajectories that are more common or specific to females 29,32 . Despite the increased inclusion of females in trajectory research, studies to date do not typically consider females specifically and have limited sample sizes and follow-up times. Persistent female offending is a rare outcome, and studies with large sample sizes and follow-ups into adulthood are needed to further enhance our understanding of this group. Follow-ups into adulthood are also needed to investigate the presence of late-onset groups. Evidence using data from England is also rare; given international differences in justice system responses, robust evidence from individual countries is needed. As females experience different drivers of offending and distinct CJS responses and outcomes compared to males 33,34 , it is important that we develop our understanding of offending pathways in females. Understanding the nature of these pathways allows us to develop targeted and time-specific interventions that can address the particular needs of offending females. This study aims to address a gap in the current literature by exploring female offending trajectories from childhood to adulthood among females who were born between 1 st September 1990 and 31 st August 1997 with at least one recorded conviction or caution between January 2000 and December 2020. This study uses a large administrative linked crime and education dataset from England and a finite mixture modelling approach to explore whether distinct latent groups exist among female offenders. We hypothesised that life-course-persistent, adolescent-limited and adult-onset groups would be identified. Methods We used STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) and REporting of studies Conducted using Observational Routinely-collected Data (RECORD) recommendations for linked cohort data for the reporting of this paper 35,36 . Ethical approval was obtained from King’s College London in December 2023 (LRS-23/24-40150). Data access was granted by the data owners (Ministry of Justice [MoJ] and Department for Education [DfE]) in May 2024. Data cleaning and analysis was performed via the Office for National Statistics Secure Research Service (ONS SRS) using R Studio and MPlus 37-39 and coding files will be made available on the corresponding author’s GitHub account upon publication (ChristabelColes). Data and Linkage The study used a longitudinal birth cohort established using linked administrative crime and education datasets, with crime data derived from the Police National Computer (PNC) and demographic data (e.g. gender, date of birth, ethnicity) derived from the National Pupil Database (NPD). The PNC contains individual offence-level records of cautions and convictions for all individuals in the UK from age 10 years onwards (the age of criminal responsibility in England, Wales and Northern Ireland). The NPD contains education and social care records for all individuals in state education in England, as well as records for national assessments taken by individuals not in state education. Self- or parent-reported gender and ethnicity variables were taken from the NPD because the equivalent records in the PNC are recorded by the arresting police officer 40 . Date of birth from the NPD was used to confirm age of offence from the PNC. PNC and NPD data were linked in 2019 as part of the MoJ and Administrative Data Research (ADR) UK funded data linkage program, Data First. Linkage was performed using a probabilistic approach agreed between the MoJ and DfE 41 . The final match rate was around 77%. The sample therefore does not capture the full offender population but has been shown to be representative of it 42 . Sample The sample consisted of females born between 1 st September 1990 and 31 st August 1997, who had at least one recorded conviction or caution in the PNC from January 2000 to December 2020. Only individuals with a record in both the NPD census and PNC were used so that we could confirm gender and age (see supplementary materials [Supplementary Figure S1] for details on the data cleaning process). After data cleaning, the final sample consisted of 196,089 females with a conviction or caution who had committed a total of 966,481 offences. Descriptive characteristics were obtained from the DfE school census and included ethnicity, free school meal (FSM) eligibility and the income deprivation affecting children index (IDACI). In line with DfE data cleaning processes 43 , where an individual’s ethnicity varied over time, the most recent record was used. FSM eligibility is determined by parental income and receipt of certain benefits (with children from lower income families being eligible) and was created by combining records from school years to create an ‘ever eligible’ variable. IDACI scores refer to the proportion of children in an area living in an income-deprived household 44 . Both measures are commonly used in research as proxy measures for socioeconomic disadvantage 45 . As IDACI scores varied over time due to familial housing movements, the most recent score was used. Home Office (HO) offence group categories were used to describe the types of offences committed by the whole sample. Additional offence characteristics used to describe offending within trajectory groups were: ever being incarcerated (according to PNC disposal codes), total length of time an individual spent in custody, serious violent offending (based on the DfE MoJ definition of serious violence offending as robbery, possession of weapons and violence against the person offences 43 ), criminal career duration (years between first and last offence), median total number of offences and ever receiving a custodial sentence of under 12 months (as short sentences make up 63% of sentences given to females and are associated with increased rates of reoffending compared to community sentences 6,46 ). For individuals born in 1990, offending information was available up until age 31 years. For those born in 1997, offending information was available up until age 24 years. See Supplementary Table S1 for a breakdown of the number of pupils in each academic year cohort. Model variables The following variables were created using offence age, offence type, offence date, HO offence codes and disposal codes from the PNC, and year of birth, month of birth and gender from the NPD. Age at first offence Individuals with missing or incorrect age in the PNC (age 31 years) were removed (N=6,144 [0.9%]). Age was then verified using year and month of birth in the NPD and individuals with inconsistent age records were removed (N=3,598 [0.5%]). Age at first offence was calculated by taking the earliest recorded offence age for each individual in the PNC. We then created a categorical age at first offence variable, with the following categories: Child (10-13 years), Adolescent (14-17 years), Young adult (18-20 years) and Adult (³21 years). These age categories correspond with previous research on offending trajectories in a mixed-gender sample using this linked data 24 . These age categories also align with research that defines developmental age groups 47,48 and categories used by the MoJ to define prolific offenders 49 . Age at last offence Age at last offence was created by taking the latest recorded offence age for each individual. We then created a categorical age at last offence variable with the following age categories: Juvenile (10-17 years), Young adult (18-20 years) and Adult (³21 years). These are the same age categories used by the MoJ to define prolific offenders 49 . Offence history We created a series of persistent offender variables which indicate whether an individual had committed more than the median number of offences (excluding zero) as a juvenile, young adult and adult. The median total number of offences committed by females from age 10-17 years was one, therefore a persistent juvenile offender was one who had committed two or more offences as a juvenile. A persistent young adult female offender was defined as an individual who had committed three or more offences from age 18-20 years. A persistent adult female offender was defined as an individual who had committed five or more offences from age 21 years onwards. This allows an individual to be persistent in one age category but not persistent in another, or persistent in all age categories. This definition was used for a previous publication using this data 24 . It is distinct from the MoJ’s 49 definition of a prolific offender, in which the categories are mutually exclusive. Violent offending A violent offender was classified as an individual who had ever committed an offence defined as ‘Violence against the person’ or a contact sexual offence (according to HO offence codes). Model covariates Custodial sentence Using PNC Disposal codes, a binary variable was created to indicate whether an individual had ever received a custodial sentence (not including suspended custodial sentences). This was used as a covariate, as individuals are largely unable to offend during periods of incarceration. This may mimic desistance and affect latent class membership, particularly for persistent offenders who are more likely to receive custodial sentences 50 . Statistical analysis Latent class analysis (LCA) was employed to model female offending trajectories. LCA is a type of finite mixture modelling used to identify homogenous groups within heterogenous populations 51 . It uses patterns of observed indicators to identify unobserved or latent groups. To prevent the model converging on local rather than global maxima, the number of random starts was set to 1000, and the number of final stage optimizations was set to 100. A maximum of 50 iterations were allowed in the initial stage. The following steps were performed to select the model that best fit the data 52 . A one-class (k) model was created, followed by models with k+1 latent classes, until models no longer converged. Log-likelihood, Bayesian information criteria (BIC), sample size adjusted BIC (aBIC) and Akaike information criteria (AIC) were recorded to compare model fit with subsequent models. The potential best-fitting models were then compared on these indicators. Models with the lowest BIC, aBIC and AIC values were favoured. These indicators give a measure of how well the model fits the data based on the penalized log-likelihood. Additionally, parsimony, class sizes, entropy and interpretability of classes within the context of previous literature were considered 53 . Posterior class probabilities were estimated and used to assign individuals to optimal groups based on the highest probability. Average posterior probabilities (AvePP) and odds of correct classification [OCC] are reported for classes in the final model. Values of over 0.8 for AvePP and 5 for OCC are considered good 54,55 . The variables described above, age at first offence, age at last offence and persistence as a juvenile, young adult and adult were used as indicator variables for all latent class models. Models were developed with and without the violent offender indicator variable, to assess whether this impacted the model. When the optimum number of classes was identified, the custodial sentence variable was added to the model as a covariate 56 . A sensitivity analysis was performed, rerunning the final 4-class model, but including individuals whose age was not verified in the NPD (due to missing or incorrect data), in order to assess the impact of excluding these individuals on the final model. Results The demographic characteristics and offence characteristics of the final sample of female offenders are presented in Tables 1 and 2. Most individuals had their first interaction with the CJS as juveniles (age 10-17 years). This was also the most common age period to be classified as a persistent offender. Most individuals had never received a custodial sentence (96.1%) or committed a violent or contact sexual offence (86.9%). The most common offence type was summary offences excluding motoring (such as common assault and battery, betting and gambling, and television license offences) (40.0%), followed by theft offences (25.5%) (Table 2). Just 3.5% (N=34,084) of all offences resulted in a custodial sentence, mainly theft (31.9%) and summary offences excluding motoring (23.6%). Table 1. Offending and demographic characteristics of sample (N=196,089) N First offence: Child (10-13 years) 48,980 (25.0%) Adolescent (14-17 years) 91,699 (46.8%) Young adult (18-20 years) 29,082 (14.8%) Adult (21-31 years) 26,328 (13.4%) Last offence: Juvenile (10-17 years) 106,198 (54.2%) Young adult (18-20 years) 38,536 (19.7%) Adult (21-31 years) 51,355 (26.2%) Persistence: 2 or more offences as a Juvenile (10-17 years) 60,686 (30.9%) 3 or more offences as a Young adult (18-20 years) 25,265 (12.9%) 5 or more offences as an Adult (21-31 years) 24,235 (12.4%) Ever been incarcerated 7,662 (3.9%) Ever committed a violent or contact sexual offence 25,656 (13.1%) Ethnicity: White 167,656 (85.5%) Black 10,200 (5.2%) Asian 5,609 (2.9%) Mixed 9,282 (4.7%) Other 1,493 (0.8%) Unknown 1,849 (0.9%) Ever eligible for FSM 103,562 (52.8%) Median IDACI score 0.25 Note: Free school meals (FSM) (population baseline rates of eligibility are around 25% 57 ); Income deprivation affecting children index (IDACI) – indicates that, as children, individuals in the sample lived in areas where on average 25.0% of children aged 0-15 were living in income-deprived families (in 2019, the highest local authority IDACI score was 0.32 44 ). Table 2. Counts of offence type and offences resulting in a custodial sentence. Counts of offences (%) Counts of offences resulting in custodial sentences (%) 01 Violence against the person 64,051 (6.6) 3,949 (11.6) 02 Sexual offences 1,112 (0.1) 260 (0.8) 03 Robbery 10,501 (1.1) 1,452 (4.3) 04 Theft Offences 246,903 (25.5) 10,881 (31.9) 05 Criminal damage and arson 11,269 (1.2) 403 (1.2) 06 Drug offences 42,634 (4.4) 1,916 (5.6) 07 Possession of weapons 13,245 (1.4) 1,025 (3.0) 08 Public order offences 17,834 (1.8) 1,326 (3.9) 09 Miscellaneous crimes against society 37,430 (3.9) 3,223 (9.5) 10 Fraud offences 24,052 (2.5) 1,122 (3.3) 11 Summary offences excluding motoring 386,267 (40.0) 8,060 (23.6) 12 Summary motoring offences 109,872 (11.4) 251 (0.7) 20 Unknown offences 1,311 (0.1) 216 (0.6) Total 966,481 (100.0) 34,084 (100.0) We estimated models from one to five classes. The addition of an extra class in the five-class model did not improve interpretability in the context of the previous literature, therefore results are not reported. The model also failed to converge without significantly increasing the number of random starts. Goodness of fit statistics for the one- to four-class model are reported in Table 3. Table 3. Model fit statistics for 1 to 4 class model (n=196,089) df Log-likelihood BIC aBIC AIC Entropy Class sizes Model 1 87 -712629.2 1425356 1425331 1425274 - 100 Model 2 78 -611005.2 1222218 1222164 1222044 0.99 46/54 Model 3 69 -563364.0 1127045 1126962 1126780 0.97 54/20/26 Model 4 60 -538191.6 1076810 1076699 1076453 0.98 54/16/11/19 Note: df: degrees of freedom; BIC: Bayesian information criterion; aBIC: adjusted Bayesian information criterion; AIC: Akaike’s information criterion. Based on having the lowest BIC, AIC and aBIC values, class sizes and the interpretability of classes, a four-class model was selected as the best-fitting model. Adding the Violent offending variable to the model did not affect the number or nature of classes identified, and it was therefore not included in the final model. The classes’ interpretation and sizes remained similar after including custodial sentence as a covariate. Fit statistics for the final four-class model, controlling for having ever served a custodial sentence, are reported in Table 4. All average posterior probabilities were over 0.9. The profiles of the identified trajectories after adjusting for having served a custodial sentence can be seen below in Table 5. The ‘Life-course-persistent’ (‘LCP’) group was the smallest group identified, accounting for 11% of the sample. These individuals began offending across childhood and adolescence and continued offending into adulthood. These individuals were persistent as juveniles and showed mixed levels of persistence in young adulthood and adulthood. ‘Adolescent-limited’ offenders were the largest group, accounting for 54% of the sample. These individuals began offending as juveniles (mainly in adolescence) and had desisted when they reached adulthood. Persistence as juveniles was mixed. The ‘Young-adult-limited' group accounted for 16% of the sample. These individuals offended in young adulthood only and persistence was mixed during this period. ‘Adult-onset’ offenders accounted for 19% of the sample. These individuals began offending after the age of 21 and were mixed in terms of persistence during adulthood. Table 4. Final model classes mcap AvePP OCC Class Life-course-persistent 0.11; 22,118 0.95 144.57 Adolescent-limited 0.54; 106,198 1.00 - Young-adult-limited 0.16; 31,055 0.95 106.45 Adult-onset 0.19; 36,718 0.96 110.94 Note: mcap: max probability class assignment proportion; AvePP: Average posterior probability, OCC: odds of correct classification Table 5. Response parameter probabilities for 4 latent classes Trajectory: Life-course-persistent 11% (N=22,118) Adolescent-limited 54% (N=106,198) Young-adult-limited 16% (N=31,055) Adult-onset 19% (N=36,718) Indicator variables: First offence childhood (10-13 years) 0.43 0.34 0.03 0.04 First offence adolescence (14-17 years) 0.57 0.66 0.12 0.14 First offence young adulthood (18-20 years) 0.00 0.00 0.85 0.10 First offence adulthood (>21 years) 0.00 0.00 0.00 0.74 Last offence juvenile (10-17 years) 0.00 0.98 0.00 0.00 Last offence young adult (18-20 years) 0.36 0.02 0.96 0.00 Last offence adulthood (>21 years) 0.64 0.01 0.04 1.00 Juvenile persistence 0.95 0.37 0.00 0.00 Young adult persistence 0.46 0.00 0.44 0.05 Adult persistence 0.37 0.00 0.00 0.46 Note: %s below 0.30 or above 0.70 are presented in bold Sensitivity analysis The sensitivity analysis, including individuals whose ages could not be verified using NPD data, confirmed that excluding individuals with inconsistent ages in the NPD did not affect results (see Supplementary Table S2 for details of the sensitivity analysis model). Characteristics of offending trajectories Table 6 shows the offending characteristics of the sample according to trajectory membership. 'LCP’ offenders were responsible for the largest proportion of offences (42%) despite making up the smallest proportion of the sample. Additionally, 19% of ‘LCP’ offenders had received a custodial sentence, compared to less than 1% of ‘Adolescent-limited’ offenders. Three percent of ‘Young-adult limited’ and five percent of ‘Adult-onset’ offenders, had received a custodial sentence. On average, 'LCP’ offenders had records for 11 offences across the study period, while ‘Adult-onset’ offenders had four, ‘Young-adult limited’ had two and ‘Adolescent-limited’ had one. For a breakdown of total offences by offence type and trajectory type, see Supplementary Table S3. Table 6. Offence and education characteristics according to latent classes Trajectory: Life-course-persistent (N=22,118) Adolescent-limited (N=106,198) Young-Adult-limited (N=31,055) Adult-onset (N=36,718) Total offences committed (% of cohort offences) 403,867 (41.8) 237,823 (24.6) 108,555 (11.2) 218,836 (22.6) Ever incarcerated 4,237 (19.2%) 667 (0.6%) 933 (3.0%) 1,825 (5.0%) Received a custodial sentence under 12 months 3,282 (14.8%) 445 (0.4%) 506 (1.6%) 1,099 (3.0%) Serious violent offenders 7,780 (35.2%) 11,348 (10.7%) 2,424 (7.8%) 3,006 (8.2%) Median total offences per person (IQR) 11 (16) 1 (1) 2 (4) 4 (5) Mean age first offence (SD) 13.8 (1.7) 14.1 (1.6) 18.1 (1.9) 21.3 (3.9) Mean age last offence (SD) 22.2 (3.3) 14.6 (1.6) 19.1 (1.1) 23.7 (2.3) Median criminal career duration (IQR) 8 (6) 1 (0) 1 (0) 1 (3) Ever been eligible for free school meals (FSM) 14,936 (67.5%) 55,733 (52.5%) 15,195 (48.9%) 17,698 (48.2%) Median Income Deprivation Affecting Children (IDACI) score 0.29 0.25 0.24 0.24 Ethnicity:* White 19,049 (86.1%) 92,396 (87.0%) 25,779 (83%) 30,432 (82.9%) Black 1,105 (5.0%) 4,624 (4.6%) 2,125 (6.8%) 2,346 (6.4%) Asian 274 (1.2%) 2,821 (2.7%) 1,107 (3.6%) 1,407 (3.8%) Mixed 1,356 (6.1%) 4,710 (4.4%) 1,458 (4.7%) 1,758 (4.8%) Other 121 (0.5%) 728 (0.7%) 277 (0.9%) 367 (1.0%) *Counts do not add up to total N of latent classes due to missing data (all classes had around 1% missing ethnicity data). Percentages may not add up to 100% due to rounding. The median IDACI score in the 'LCP’ group was 0.29; meaning that, as children, these female offenders lived in neighbourhoods where on average 29% of children lived in families with low-income. The other groups had similar median IDACI scores ranging from 0.24-0.25. Additionally, 68% of individuals in the 'LCP' trajectory had received free school meals at some point whilst at school, compared to 48-52% of individuals in the other trajectories. See Table Supplementary S4 for further demographic and offence characteristics of latent classes. Discussion Females have long been neglected in research on life course patterns of offending. It is key that we look at these patterns as they may allow us to distinguish between groups of offenders with distinct risk and recidivism factors and treatment/disposal needs. We used a large administrative sample from linked PNC and NPD data to identify subgroups of female offenders from the age of criminal responsibility (age 10 years) up to age 31 years. This is the largest study to date specifically exploring female offending trajectories and the first in England to explore female-only trajectories of reoffending from childhood well into adulthood with a nationally representative sample. We used latent class analysis to identify four groups with different patterns of onset, desistance and rates of offending: Life-course-persistent, Adolescent-limited, Young-Adult-limited and Adult-onset. While these groups resemble those found in previous literature using both male and female samples 21,58 , the characteristics and associated risk factors for the identified groups are likely to differ between males and females. It is noteworthy that over 50% of the offending in this cohort was for summary offences (including motoring) and that these summary offences and theft accounted for over 76% of the offences, but that considerable serious offending was still observed. The findings contribute to our understanding of heterogeneity in female offending and demonstrate the need for developmentally and trajectory appropriate interventions. We identified an LCP trajectory, resembling Moffitt’s 17 LCP group, that accounted for 11% of the sample. These offenders made up the majority of the incarcerated population in the sample (of the 7,662 individuals who had ever been incarcerated, 55% were classified as LCP offenders). This may reflect the high average number of offences committed by this group (11 offences per offender), as previous convictions are considered in sentencing decisions 59 . Fifteen percent of LCP offenders had received custodial sentences of under 12 months. Such sanctions are commonly given to female offenders and are associated with increased reoffending rates compared to community-based sentences 6,60 , although the baseline characteristics of these two groups may differ 15 . These offenders also displayed high rates of socioeconomic disadvantage with 68% eligible for FSMs compared to around 25% in the general population 57 . As this group accounts for the largest share of offences among female offenders, they should be a priority for interventions. Early intensive interventions focusing on family 61 and school environments 62 may be beneficial for this group. Female persistent offending is more likely to be linked to the antisocial influence of partners, mental health disorders and exposure to violence compared to male persistence 26 . Qualitative perspectives also highlight the role of victimisation in driving young girls to commit particular offences (e.g. petty property offences and offences related to prostitution) 63 . Recent research modeling offending trajectories among care-experienced individuals found that females who experienced out of home placements during adolescence were at the highest risk of becoming persistent offenders 64,65 . Future research should further explore the early correlates and predictors of membership in this group. This may help to prospectively identify at-risk females for interventions in schools or other education settings, prior to offending onset. Interventions should target the gender-specific needs of high-risk females (e.g. family separation, substance misuse, and disconnection from school 66 ). This could allow us to more effectively distribute resources to offenders at higher risk of negative life outcomes. This may help to prevent the onset of these high-rate criminal careers, reducing the associated individual, societal, and inter-generational costs. The Adolescent-limited offending group identified only committed offences as juveniles and made up around 54% of the total sample. This indicates that, in line with Moffitt’s theory, this is the most common pattern of offending amongst females 67 . It is unclear why these offenders desist, but this may be due to a narrowing of the gap between social and biological maturity over time 17 . Qualitative work exploring this adolescent group has identified delinquent peers and identity exploration as important factors for offending in this group 68 . While this group appear to have significantly better later life outcomes than persistent groups 69 , they do show deficits in other areas, such as educational attainment 70 , indicating a need for support. The Young-adult-limited group identified in this study made up 16% of the total sample. This group has recently been identified in male offender samples 64,65 . They had higher rates of incarceration (3%) than Adolescent-limited offenders (0.6%). This may be due to differences in sentencing guidelines between adults and minors. Similar to the Adolescent-limited group, these offenders appeared to age out of offending as they matured. This group may have more protective factors than their persistent counterparts, e.g. active employment 71 . Previous research on female offenders has found that this group had similar risk factors to very low-level sporadic offenders 71 . They also experienced different outcomes later in life; for example, the majority of late-onset desisting offenders who married later experienced a divorce. Late adolescence may be a critical window for interventions in this group. Due to their age, it is key that this group does not fall in the gap between the rehabilitation-oriented youth justice system and the more punitive adult justice system Services that support women in this transitional age period (e.g. with employment, housing, vocational skills) may help to prevent the onset of offending for these individuals. Adult-onset groups have caused disagreements in previous literature, with some research maintaining that this group is an artifact of methodological factors and unidentified juvenile offences 18 . In our sample this group made up 19% of offenders. The late onset of this group indicates the need to investigate adulthood risk factors for offending onset in females, and to potentially target specific interventions later in adolescence or early adulthood. These offenders may represent ‘late-bloomers’ 72 , a group who start offending during adulthood but reach the same level of offending as LCP offenders in later life. However, previous research also finds adult-onset groups that desist over time 71 . As our sample was only followed-up until age 31 years, it is unclear whether this group persists or desists throughout later adulthood. The later-onset of this group (at or after 21 years) may be explained by the presence of coping strategies and support networks in adolescence and early adulthood, that stop being effective as they experience the reduced social support and increasing demands that come with later adulthood 73 . Limitations Finite mixture modelling (FMM) approaches, including latent class analysis, have typically been used to examine longitudinal patterns of offending 19 . Criticisms of FMM include the potential misinterpretation of identified groups as ‘real’ subgroups 74 . It should be acknowledged that groups identified using FMM are not real entities but probabilistic approximations of patterns in outcomes and should therefore be interpreted accordingly. Administrative offending records only account for offences that have been dealt with by the CJS. Offending rates are likely to be underestimated, particularly for young female offenders, who may be less likely to receive official sanctions due to committing less serious offences 75 . It is estimated that the actual age of offending onset is around 3-5 years prior to an individual’s first recorded offence 76 . Official measures may also be a proxy for more serious offences as ‘normative’ adolescent offending may not be captured in these records 72 . The maximum follow-up in this study was to age 31 years. While this is an improvement on much previous research (most studies that continue into adulthood do not follow-up past age 26), ideally follow-up would be further into adulthood. As the birth cohort used in this research ages, this information will become available for further analysis. Additionally, we have an uneven length of follow-up for individuals due to the seven-year span of the birth cohort (1990-1997). This means that our study end-point ranged from 24 to 31 years of age. This has the potential to influence latent class assignment: however, individuals in each birth year are relatively evenly distributed between identified trajectories (see Supplementary Table S4). We could not control for mortality or migration, as neither the NPD nor PNC contain information on this. This may impact the groups identified, because if someone dies or emigrates, they will appear to have desisted from offending. Premature aging and early death are more common amongst persistent offenders 77,78 , therefore this may have resulted in an under-identification of LCP offenders. Future Directions The current study exemplifies the power of using administrative data to answer questions related to public policy, using near whole-population samples with relatively low financial and time burdens. Recent work using UK administrative data, for example looking at changes in educational attainment and criminal offending 79 and the intersection between care experience, ethnicity and offending 80 , shows the potential of using linked datasets to explore social justice issues. Further research on the social care and educational backgrounds of female offenders, particularly the LCP group identified, would help to guide research focused on gender-informed approaches that incorporate a Whole Systems Approach model from childhood onwards 81 . Conclusion This study has enhanced our theoretical understanding of female offending trajectories through harnessing a large administrative sample of female offenders convicted or cautioned for an offence in England between 2000 and 2020. This is the largest study looking at female-only trajectories to date. By using a large nationally-representative dataset we move beyond small-scale studies and provide a foundation for further gender-informed criminal justice policy research. We identified four distinct trajectories of offending among females and explored offence and some demographic characteristics of these groups. Our findings highlight the need for gender-responsive strategies that account for distinct female offending trajectories. Through developing our theoretical understanding of female offending patterns, we can inform criminal justice interventions to support and rehabilitate female offenders. Declarations Funding CC received support for this work from Administrative Data Research (ADR) UK, an Economic and Social Research Council (ESRC) investment (Grant number: ES/P000703/1). Acknowledgements This work was undertaken in the Office for National Statistics Secure Research Service using data from the ONS and other owners and does not imply the endorsement of the ONS or other data owners. Author contributions Authors HD, AW and CC contributed to the article conceptualisation. CC performed the data cleaning and analysis and prepared the original draft. CC was the only author to have access to the dataset. Authors AW, DJ, NB, and HD critically reviewed and edited the work. Data availability statement Data used for this research is accessible via an application by an ONS accredited researcher to the data owners (Ministry of Justice and Department for Education). Additional Information (including a Competing Interests Statement) AW and NB are supported by the National Institute for Health and Care Research (NIHR) Maudsley Biomedical Research Centre (BRC). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care. References Fair, H. & Walmsley, R. World Female Imprisonment List. (Institute for Crime and Justice Policy Research (ICPR), London, UK, 2022). Favril, L., Rich, J. D., Hard, J. & Fazel, S. Mental and physical health morbidity among people in prisons: an umbrella review. The Lancet. Public health 9 , e250-e260 (2024). https://doi.org/10.1016/S2468-2667(24)00023-9 McLeod, K. E. et al. Health conditions among women in prisons: a systematic review. The Lancet. Public health 10 , e609-e624 (2025). https://doi.org/10.1016/S2468-2667(25)00092-1 Nilsson, S. F., Nordentoft, M., Fazel, S. & Laursen, T. M. Risk of homelessness after prison release and recidivism in Denmark: a nationwide, register-based cohort study. The Lancet. Public health 8 , e756-e765 (2023). https://doi.org/10.1016/S2468-2667(23)00152-4 Karlsson, M. E. & Zielinski, M. J. Sexual Victimization and Mental Illness Prevalence Rates Among Incarcerated Women: A Literature Review. Trauma, violence & abuse 21 , 326-349 (2020). https://doi.org/10.1177/1524838018767933 Ministry of Justice. Female Offender Strategy. (HMSO, London, 2018). Kelman, J., Gribble, R., Harvey, J., Palmer, L. & MacManus, D. How Does a History of Trauma Affect the Experience of Imprisonment for Individuals in Women's Prisons: A Qualitative Exploration. Women & criminal justice 34 , 171-191 (2024). https://doi.org/10.1080/08974454.2022.2071376 Ministry of Justice. Safety in custody quarterly: update to December 2017. (2018). Prison Reform Trust. Resetting the approach to women’s imprisonment , (2025). Lee, R. D., Fang, X. & Luo, F. The impact of parental incarceration on the physical and mental health of young adults. Pediatrics (Evanston) 131 , e1188-e1195 (2013). https://doi.org/https://doi.org/10.1542/peds.2012-0627 Murray, J., Farrington, D. P. & Sekol, I. Children's Antisocial Behavior, Mental Health, Drug Use, and Educational Performance After Parental Incarceration: A Systematic Review and Meta-Analysis. Psychological bulletin 138 , 175-210 (2012). https://doi.org/https://doi.org/10.1037/a0026407 Poehlmann, J. Representations of Attachment Relationships in Children of Incarcerated Mothers. Child development 76 , 679-696 (2005). https://doi.org/10.1111/j.1467-8624.2005.00871.x Gelsthorpe, L. in Women who offend (ed Gill McIvor) (Jessica Kingsley Publishers, 2004). Steffensmeier, D. & Allan, E. Gender and Crime: Toward a Gendered Theory of Female Offending. Annual review of sociology 22 , 459-487 (1996). https://doi.org/10.1146/annurev.soc.22.1.459 Yukhnenko, D., Wolf, A., Blackwood, N. & Fazel, S. Recidivism rates in individuals receiving community sentences: A systematic review. PloS one 14 , e0222495 (2019). https://doi.org/10.1371/journal.pone.0222495 Piquero, A. R. “We study the past to understand the present; we understand the present to guide the future”: The time capsule of developmental and life-course criminology. Journal of criminal justice 85 , 101932 (2023). https://doi.org/10.1016/j.jcrimjus.2022.101932 Moffitt, T. E. Adolescence-Limited and Life-Course-Persistent Antisocial Behavior: A Developmental Taxonomy. Psychological review 100 , 674-701 (1993). https://doi.org/https://doi.org/10.1037/0033-295X.100.4.674 Moffitt, T. E. Male antisocial behaviour in adolescence and beyond. Nature human behaviour 2 , 177-186 (2018). https://doi.org/https://doi.org/10.1038/s41562-018-0309-4 Nagin, D. S. & Land, K. C. AGE, CRIMINAL CAREERS, AND POPULATION HETEROGENEITY: SPECIFICATION AND ESTIMATION OF A NONPARAMETRIC, MIXED POISSON MODEL. Criminology (Beverly Hills) 31 , 327-362 (1993). https://doi.org/https://doi.org/10.1111/j.1745-9125.1993.tb01133.x Silverthorn, P., Frick, P. J. & Reynolds, R. Timing of onset and correlates of severe conduct problems in adjudicated girls and boys. Journal of Psychopathology and Behavioral Assessment 23 , 171-181 (2001). https://doi.org/https://doi.org/10.1023/A:1010917304587 Jennings, W. G. & Reingle, J. M. On the number and shape of developmental/life-course violence, aggression, and delinquency trajectories: A state-of-the-art review. Journal of criminal justice 40 , 472-489 (2012). https://doi.org/https://doi.org/10.1016/j.jcrimjus.2012.07.001 Jolliffe, D., Farrington, D. P., Piquero, A. R., MacLeod, J. F. & van de Weijer, S. Prevalence of life-course-persistent, adolescence-limited, and late-onset offenders: A systematic review of prospective longitudinal studies. Aggression and violent behavior 33 , 4-14 (2017). https://doi.org/10.1016/j.avb.2017.01.002 Jolliffe, D., Farrington, D. P., Piquero, A. R., Loeber, R. & Hill, K. G. Systematic review of early risk factors for life-course-persistent, adolescence-limited, and late-onset offenders in prospective longitudinal studies. Aggression and violent behavior 33 , 15-23 (2017). https://doi.org/10.1016/j.avb.2017.01.009 Dickson, H. Childhood educational predictors of reoffending trajectories , (2024). Gushue, K., McCuish, E. C. & Corrado, R. R. Developmental Offending Patterns: Female Offending Beyond the Reference Category. Criminal justice and behavior 48 , 139-156 (2021). https://doi.org/10.1177/0093854820966729 Cauffman, E., Monahan, K. C. & Thomas April, G. Pathways to Persistence: Female Offending from 14 to 25. Journal of Developmental and Life-Course Criminology 1 , 236-268 (2015). https://doi.org/https://doi.org/10.1007/s40865-015-0016-z Odgers, C. L. et al. Female and male antisocial trajectories: From childhood origins to adult outcomes. Development and psychopathology 20 , 673-716 (2008). https://doi.org/https://doi.org/10.1017/S0954579408000333 Sivertsson, F., Carlsson, C., Almquist, Y. B. & Brännström, L. Offending trajectories from childhood to retirement age: Findings from the Stockholm birth cohort study. Journal of criminal justice 91 , 102155 (2024). https://doi.org/https://doi.org/10.1016/j.jcrimjus.2024.102155 Andersson, F., Levander, S., Svensson, R. & Levander, M. T. Sex differences in offending trajectories in a Swedish cohort: Sex differences in offending trajectories. Criminal behaviour and mental health 22 , 108-121 (2012). https://doi.org/https://doi.org/10.1002/cbm.1822 Carlsson, C. & Sivertsson, F. Age, gender, and crime in a Stockholm birth cohort to age 64. Journal of Developmental and Life-Course Criminology 7 , 359-384 (2021). Fergusson, D. M. & Horwood, L. J. Male and female offending trajectories. Development and psychopathology 14 , 159-177 (2002). https://doi.org/https://doi.org/10.1017/S0954579402001098 Kuluk, A., Allard, T., Thompson, C., Ogilvie, J. M. & Broidy, L. Offending Trajectories in an Australian Birth Cohort: Differences and Similarities Across Sex. Criminal justice and behavior 51 , 807-830 (2024). https://doi.org/10.1177/00938548241234373 Cauffman, E. Understanding the Female Offender. The Future of children 18 , 119-142 (2008). https://doi.org/10.1353/foc.0.0015 Ministry of Justice. Statistics on Women and the Criminal Justice System 2023. (2025). Benchimol, E. I. et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD). PLoS Med (2015). von Elm, E. et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. Epidemiology (Cambridge, Mass.) 18 , 800-804 (2007). https://doi.org/10.1097/EDE.0b013e3181577654 Mplus User’s Guide (Muthén & Muthén, Los Angeles, CA, 1998-2017). R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, Austria, 2024). RStudio: Integrated Development Environment for R (Posit Software, PBC, Boston, MA, 2024). Wickersham, A. et al. Educational attainment trajectories among children and adolescents with depression, and the role of sociodemographic characteristics: longitudinal data-linkage study. British journal of psychiatry 218 , 151-157 (2021). https://doi.org/https://doi.org/10.1192/bjp.2020.160 Ministry of Justice. A Technical Guide to Statistics on Race and the Criminal Justice System. (2018). Department for Education. Education, children’s social care and offending: multi-level modelling: technical report., (2023). Department for Education and Ministry of Justice. Education, children’s social care and offending. (2022). Ministry of Housing, C. L. G. The English Indices of Deprivation 2019 (IoD2019). (2019). Crawford, C. & Greaves, E. A comparison of commonly used socio-economic indicators: their relationship to educational disadvantage and relevance to Teach First. (Institute for Fiscal Studies, 2013). Prison Reform Trust. Why focus on reducing women’s imprisonment? , (2022). Bodker, A. & Harris, D. A. Help-Seeking Behaviors of Young People with Problematic Sexualized Behaviors. Journal of child sexual abuse , 1-20 (2024). https://doi.org/10.1080/10538712.2024.2415550 Sampson, R. J. & Laub, J. H. LIFE-COURSE DESISTERS? TRAJECTORIES OF CRIME AMONG DELINQUENT BOYS FOLLOWED TO AGE 70*. Criminology 41 , 555 (2003). https://doi.org/10.1111/j.1745-9125.2003.tb00997.x Ministry of Justice. Prolific Offenders - Defining a Prolific Offender. (2017). Farrington, D. P. & West, D. J. Criminal, penal and life histories of chronic offenders: risk and protective factors and early identification. Criminal behaviour and mental health 3 , 492-523 (1993). https://doi.org/10.1002/cbm.1993.3.4.492 Herle, M. et al. Identifying typical trajectories in longitudinal data: modelling strategies and interpretations. European journal of epidemiology 35 , 205-222 (2020). https://doi.org/10.1007/s10654-020-00615-6 Masyn, K. E. in The Oxford Handbook of Quantitative Methods (ed T. D. Little) (Oxford University Press, 2013). Nagin, D. Group-based modeling of development . 1st edn, (Harvard University Press, 2005). Nagin, D. S. & Odgers, C. L. Group-Based Trajectory Modeling (Nearly) Two Decades Later. Journal of Quantitative Criminology 26 , 445 (2010). https://doi.org/https://doi.org/10.1007/s10940-010-9113-7 Clark, S. & Muthén, B. Relating Latent Class Analysis Results to Variables not Included in the Analysis. (2009). Vermunt, J. K. Latent Class Modeling with Covariates: Two Improved Three-Step Approaches. Political analysis 18 , 450-469 (2010). https://doi.org/10.1093/pan/mpq025 Department for Education. Schools, pupils and their characteristics. (2024). Fontaine, N., Carbonneau, R., Vitaro, F., Barker, E. D. & Tremblay, R. E. Research review: A critical review of studies on the developmental trajectories of antisocial behavior in females. Journal of child psychology and psychiatry 50 , 363-385 (2009). https://doi.org/10.1111/j.1469-7610.2008.01949.x Sentencing Council Guidelines. Imposition of community and custodial sentences - Effective from 1 April 2025. (2025). Prison Reform Trust. Six in 10 women sent to prison serve sentences of less than six months , (2023). Leijten, P. et al. Research Review: Harnessing the power of individual participant data in a meta‐analysis of the benefits and harms of the Incredible Years parenting program. Journal of child psychology and psychiatry 59 , 99-109 (2018). https://doi.org/10.1111/jcpp.12781 Veenman, B., Luman, M. & Oosterlaan, J. Efficacy of behavioral classroom programs in primary school. A meta-analysis focusing on randomized controlled trials. PloS one 13 , e0201779 (2018). https://doi.org/10.1371/journal.pone.0201779 Chesney-Lind, M. & Shelden, R. G. Girls, delinquency, and juvenile justice . Fourth edition. edn, (Wiley, 2014). Brännström, L., Andershed, A.-K., Vinnerljung, B., Hjern, A. & Almquist, Y. B. Gender-specific trajectories of offending from adolescence until age 40 among individuals with experience of out-of-home care: A national cohort study. Developmental child welfare 6 , 3-22 (2024). https://doi.org/10.1177/25161032231217265 Hossein, G. A prospective examination of offending trajectories in men and women with out-of-home care experience followed up until retirement age. Child Abuse & Neglect 161 , 107253 (2025). https://doi.org/https://doi.org/10.1016/j.chiabu.2025.107253 Shepherd, S. M., Newton, D., Harries, C., Fix, R. L. & Fullam, R. An analysis of high-risk offending pathways for young females in custody. Psychiatry, psychology, and law 26 , 194-205 (2019). https://doi.org/10.1080/13218719.2018.1487344 Moffitt, T. E. & Caspi, A. Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. Development and Psychopathology 13 , 355-375 (2001). https://doi.org/https://doi.org/10.1017/S0954579401002097 Azad, A. & Carlsson, J. Identity status and narrative identity processes in female adolescents' stories about committing crimes and being convicted. Journal of adolescence (London, England.) 96 , 124-135 (2024). https://doi.org/10.1002/jad.12261 Pulkkinen, L., Fadjukoff, P. & Pitkänen, T. Persistent offenders and adolescence‐limited offenders: Differences in life‐courses. Criminal Behaviour and Mental Health : CBMH 30 , 196-209 (2020). https://doi.org/https://doi.org/10.1002/cbm.2157 Azad, A. & Ginner Hau, H. Adolescent Females with Limited Delinquency: A Follow-Up on Educational Attainment and Recidivism. Child & youth care forum 49 , 325-342 (2020). https://doi.org/10.1007/s10566-019-09530-8 Blokland, A. & van Os, R. Life Span Offending Trajectories of Convicted Dutch Women. International criminal justice review 20 , 169-187 (2010). https://doi.org/10.1177/1057567710368938 Matsuda, M., Thornberry, T. P., Loughran, T. A. & Krohn, M. D. Are Late Bloomers Real? Identification and Comparison of Late-Onset Offending Patterns from Ages 14–40. Journal of Developmental and Life-Course Criminology 8 , 124-150 (2022). https://doi.org/https://doi.org/10.1007/s40865-022-00189-9 Broidy, L., Payne, J. & Piquero, A. R. Making Sense of Heterogeneity in the Influence of Childhood Abuse, Mental Health, and Drug Use on Women’s Offending Pathways. Criminal Justice and Behavior 45 , 1565-1587 (2018). https://doi.org/https://doi.org/10.1177/0093854818776687 Skardhamar, T. DISTINGUISHING FACTS AND ARTIFACTS IN GROUP-BASED MODELING. Criminology (Beverly Hills) 48 , 295-320 (2010). https://doi.org/https://doi.org/10.1111/j.1745-9125.2010.00185.x D'Unger, A. V., Land, K. C. & McCall, P. L. Sex Differences in Age Patterns of Delinquent/Criminal Careers: Results from Poisson Latent Class Analyses of the Philadelphia Cohort Study. Journal of quantitative criminology 18 , 349-375 (2002). https://doi.org/10.1023/A:1021117626767 Theobald, D. & Farrington, D. P. in Encyclopedia of Criminology and Criminal Justice (ed G. Bruinsma, Weisburd, D.) (Springer, 2014). Langevin, S. et al. Life-Course Persistent Antisocial Behavior and Accelerated Biological Aging in a Longitudinal Birth Cohort. International journal of environmental research and public health 19 , 14402 (2022). https://doi.org/10.3390/ijerph192114402 Zane, S. N., Welsh, B. C. & Zimmerman, G. M. Criminal Offending and Mortality over the Full Life-Course: A 70-Year Follow-up of the Cambridge–Somerville Youth Study. Journal of Quantitative Criminology 35 , 691-713 (2019). https://doi.org/https://doi.org/10.1007/s10940-018-9399-4 Wickersham, A. Changes in school performance and involvement in the criminal justice system , (2024). Hunter, K. The time is now: Preventing unnecessary criminalisation of children in care. International journal of population data science 8 (2023). https://doi.org/10.23889/ijpds.v8i2.2193 Ministry of Justice. Female Offending Blueprint for Wales. (2019). Additional Declarations Competing interest reported. AW and NB are supported by the National Institute for Health and Care Research (NIHR) Maudsley Biomedical Research Centre (BRC). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care. Supplementary Files Supplementarymaterials.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8115965","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":548220352,"identity":"e9a25321-e24b-4e2f-a478-f12ccf89c720","order_by":0,"name":"Christabel Coles","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABE0lEQVRIie2RsUrEQBCGZ1k4m9Vtk8ZnGDiwOpJX2WUh5SFcIygoBM4mahvwJVLZOmHhrslxrYXF3RvERmKjbu4KFZJoabHfFsswfDv/sAAez38kACB3KThoq1MAKYDRriWGFOUObysECLP9I4MK/FCw+kWR9zfb8q2JppIDq2ucxOPKbgguIudS95DnJVqhzCxMgYc5JvphNUeChQFcXXUqGCRgQXFdWBhxgVadrIULNiLAdXewVikbdakf98pHPL5rlfdhhYSyuuA7hVhxmAGxOfUGC55cMJEsZ4FlqdvF6LxaIOlbI8Ke9WWe8Jdmcj6V12lZ12dRLDOz3dSv0fFRpbqTfcG+5VD9H+nxeDyeP/AJRTJcZa/3bi4AAAAASUVORK5CYII=","orcid":"","institution":"King's College London","correspondingAuthor":true,"prefix":"","firstName":"Christabel","middleName":"","lastName":"Coles","suffix":""},{"id":548220353,"identity":"93471c45-d558-4f0a-968a-d0b3d3507193","order_by":1,"name":"Alice Wickersham","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"prefix":"","firstName":"Alice","middleName":"","lastName":"Wickersham","suffix":""},{"id":548220354,"identity":"41d4c3ed-9ed3-4f72-8cae-4f4a3065fe14","order_by":2,"name":"Darrick Jolliffe","email":"","orcid":"","institution":"University College London","correspondingAuthor":false,"prefix":"","firstName":"Darrick","middleName":"","lastName":"Jolliffe","suffix":""},{"id":548220355,"identity":"eda7bac2-6e6d-4ad5-99a9-d6a07cf71145","order_by":3,"name":"Nigel Blackwood","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"prefix":"","firstName":"Nigel","middleName":"","lastName":"Blackwood","suffix":""},{"id":548220356,"identity":"897c1575-ed09-4317-a64e-dae90e48b090","order_by":4,"name":"Hannah Dickson","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"prefix":"","firstName":"Hannah","middleName":"","lastName":"Dickson","suffix":""}],"badges":[],"createdAt":"2025-11-14 14:23:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8115965/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8115965/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96711596,"identity":"f6a8e0d5-9ddd-4da5-a893-0cfb5e5fa0e3","added_by":"auto","created_at":"2025-11-25 10:12:17","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1008728,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8115965/v1/fcc349f7-8009-49e3-982b-38841b6b596e.pdf"},{"id":96693580,"identity":"a5b027de-0cec-43b0-90db-dafa01742e9c","added_by":"auto","created_at":"2025-11-25 07:13:36","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":39800,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-8115965/v1/350ff88d1b089ecafc22ffca.docx"}],"financialInterests":"Competing interest reported. AW and NB are supported by the National Institute for Health and Care Research (NIHR) Maudsley Biomedical Research Centre (BRC). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.","formattedTitle":"Female reoffending trajectories in England from 2000 to 2020: A longitudinal administrative data study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe UK has one of the highest female imprisonment rates in Western Europe \u003csup\u003e1\u003c/sup\u003e. Women who are imprisoned have a significantly elevated prevalence of mental disorder, substance misuse, homelessness and traumatic life experiences \u003csup\u003e2-6\u003c/sup\u003e. Imprisonment may serve as a re-traumatising experience \u003csup\u003e7\u003c/sup\u003e, with female prisoners\u0026rsquo; rates of self-harm nearly five times higher than males\u0026rsquo; \u003csup\u003e8\u003c/sup\u003e. In addition, in 2020, over 17,500 children were estimated to have been separated from their mother as a result of maternal imprisonment \u003csup\u003e9\u003c/sup\u003e. Such separations are associated with insecurities of attachment, depression and an increased risk of antisocial behaviour \u003csup\u003e10-12\u003c/sup\u003e. Despite the notable differences and impacts there has been little research about female offending and this is particularly true in England and Wales \u003csup\u003e13\u003c/sup\u003e. This may be because women tend to make up a small proportion of offenders, and their offences are on average less serious \u003csup\u003e14\u003c/sup\u003e, but the result of this absence is that most criminological theories and indeed criminal justice approaches are simply adapted from research and knowledge about males. We must therefore enhance our understanding of the developmental trajectories of female reoffending and their associated risk factors to inform prevention and intervention strategies, including reduced imprisonment and increased community-based sentencing strategies in this group \u003csup\u003e15\u003c/sup\u003e. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDevelopmental trajectories of offending explore the antecedents and course of offending behaviours across the lifespan \u003csup\u003e16\u003c/sup\u003e. Moffitt drew on developmental trajectory work to suggest that two main offending trajectories exist: life-course-persistent (LCP) and adolescent-limited (AL) \u003csup\u003e17\u003c/sup\u003e. LCP offending is characterised by an early onset of serious and diverse offending that persists across the life course, associated with a combination of neurodevelopmental and environmental risk factors. AL offending is characterised by less serious offending, mainly during the adolescent years, associated with a delinquent peer group and a gap between biological and social maturity. While both groups of offenders were predominantly male, female offenders were more typically observed in the adolescent-limited group. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSince Moffitt\u0026rsquo;s publication, life course trajectories of male offending from childhood into adulthood and their correlates have been well researched \u003csup\u003e18\u003c/sup\u003e. Methods used to identify groups have varied, with some studies using finite mixture modelling approaches \u003csup\u003e19\u003c/sup\u003e and others using hard-coding methods and algorithms \u003csup\u003e20\u003c/sup\u003e. Finite mixture modelling is a data-driven approach that estimates groups probabilistically, whereas hard-coding and algorithm methods use pre-defined rules or thresholds and do not account for population heterogeneity. Such work has cumulatively suggested that additional trajectories beyond LCP and AL offending exist, including late-onset escalating and early-onset desistant trajectories \u003csup\u003e21\u003c/sup\u003e. The exact number of groups identified appears to be moderated by methodological factors: for example, studies which used self-report rather than official data identified more offending trajectories \u003csup\u003e21\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo date, most research exploring offending trajectories has been performed using male or mixed-gender samples (see \u003csup\u003e21-23\u003c/sup\u003e for reviews). Recent work using a mixed-gender birth cohort from the linked Police National Computer and National Pupil Database identified five offending trajectories (including an LCP, two ALs, and two late-onset) \u003csup\u003e24\u003c/sup\u003e. However, these findings are likely to have been driven by the large proportion of male offenders in the sample. True heterogeneity in female re-offending patterns and characteristics of trajectories may be masked in such studies due to higher rates of male offending \u003csup\u003e25\u003c/sup\u003e. Overall, research on female trajectories has found similar trajectory types to those in males but has also identified some important differences. While research has found female LCP groups \u003csup\u003e26-28\u003c/sup\u003e, women began offending at a later age \u003csup\u003e29,30\u003c/sup\u003e, committed fewer violent and serious offences \u003csup\u003e31\u003c/sup\u003e and were less likely to follow persistent offending trajectories compared to men. Some studies also find adult-onset trajectories that are more common or specific to females \u003csup\u003e29,32\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDespite the increased inclusion of females in trajectory research, studies to date do not typically consider females specifically and have limited sample sizes and follow-up times. Persistent female offending is a rare outcome, and studies with large sample sizes and follow-ups into adulthood are needed to further enhance our understanding of this group. Follow-ups into adulthood are also needed to investigate the presence of late-onset groups. Evidence using data from England is also rare; given international differences in justice system responses, robust evidence from individual countries is needed. As females experience different drivers of offending and distinct CJS responses and outcomes compared to males \u003csup\u003e33,34\u003c/sup\u003e, it is important that we develop our understanding of offending pathways in females. Understanding the nature of these pathways allows us to develop targeted and time-specific interventions that can address the particular needs of offending females.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis study aims to address a gap in the current literature by exploring female offending trajectories from childhood to adulthood among females who were born between 1\u003csup\u003est\u003c/sup\u003e September 1990 and 31\u003csup\u003est\u003c/sup\u003e August 1997 with at least one recorded conviction or caution between January 2000 and December 2020. This study uses a large administrative linked crime and education dataset from England and a finite mixture modelling approach to explore whether distinct latent groups exist among female offenders. We hypothesised that life-course-persistent, adolescent-limited and adult-onset groups would be identified. \u0026nbsp; \u0026nbsp;\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eWe used STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) and REporting of studies Conducted using Observational Routinely-collected Data (RECORD) recommendations for linked cohort data for the reporting of this paper \u003csup\u003e35,36\u003c/sup\u003e. Ethical approval was obtained from King\u0026rsquo;s College London in December 2023 (LRS-23/24-40150). Data access was granted by the data owners (Ministry of Justice [MoJ] and Department for Education [DfE]) in May 2024. Data cleaning and analysis was performed via the Office for National Statistics Secure Research Service (ONS SRS) using R Studio and MPlus\u003csup\u003e37-39\u003c/sup\u003e and coding files will be made available on the corresponding author\u0026rsquo;s GitHub account upon publication (ChristabelColes).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData and Linkage\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study used a longitudinal birth cohort established using linked administrative crime and education datasets, with crime data derived from the Police National Computer (PNC) and demographic data (e.g. gender, date of birth, ethnicity) derived from the National Pupil Database (NPD).\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eThe PNC contains individual offence-level records of cautions and convictions for all individuals in the UK from age 10 years onwards (the age of criminal responsibility in England, Wales and Northern Ireland). The NPD contains education and social care records for all individuals in state education in England, as well as records for national assessments taken by individuals not in state education. Self- or parent-reported gender and ethnicity variables were taken from the NPD because the equivalent records in the PNC are recorded by the arresting police officer \u003csup\u003e40\u003c/sup\u003e. Date of birth from the NPD was used to confirm age of offence from the PNC.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePNC and NPD data were linked in 2019 as part of the MoJ and Administrative Data Research (ADR) UK funded data linkage program, Data First. Linkage was performed using a probabilistic approach agreed between the MoJ and DfE \u003csup\u003e41\u003c/sup\u003e. The final match rate was around 77%. The sample therefore does not capture the full offender population but has been shown to be representative of it \u003csup\u003e42\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSample\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe sample consisted of females born between 1\u003csup\u003est\u003c/sup\u003e September 1990 and 31\u003csup\u003est\u003c/sup\u003e August 1997, who had at least one recorded conviction or caution in the PNC from January 2000 to December 2020. Only individuals with a record in both the NPD census and PNC were used so that we could confirm gender and age (see supplementary materials [Supplementary Figure S1] for details on the data cleaning process). After data cleaning, the final sample consisted of 196,089 females with a conviction or caution who had committed a total of 966,481 offences.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDescriptive characteristics were obtained from the DfE school census and included ethnicity, free school meal (FSM) eligibility and the income deprivation affecting children index (IDACI). In line with DfE data cleaning processes \u003csup\u003e43\u003c/sup\u003e, where an individual\u0026rsquo;s ethnicity varied over time, the most recent record was used. FSM eligibility is determined by parental income and receipt of certain benefits (with children from lower income families being eligible) and was created by combining records from school years to create an \u0026lsquo;ever eligible\u0026rsquo; variable. IDACI scores refer to the proportion of children in an area living in an income-deprived household \u003csup\u003e44\u003c/sup\u003e. Both measures are commonly used in research as proxy measures for socioeconomic disadvantage \u003csup\u003e45\u003c/sup\u003e. As IDACI scores varied over time due to familial housing movements, the most recent score was used. Home Office (HO) offence group categories were used to describe the types of offences committed by the whole sample. Additional offence characteristics used to describe offending within trajectory groups were: ever being incarcerated (according to PNC disposal codes), total length of time an individual spent in custody, serious violent offending (based on the DfE MoJ definition of serious violence offending as robbery, possession of weapons and violence against the person offences \u003csup\u003e43\u003c/sup\u003e), criminal career duration (years between first and last offence), median total number of offences and ever receiving a custodial sentence of under 12 months (as short sentences make up 63% of sentences given to females and are associated with increased rates of reoffending compared to community sentences \u003csup\u003e6,46\u003c/sup\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor individuals born in 1990, offending information was available up until age 31 years. For those born in 1997, offending information was available up until age 24 years. See Supplementary Table S1 for a breakdown of the number of pupils in each academic year cohort.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel variables\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe following variables were created using offence age, offence type, offence date, HO offence codes and disposal codes from the PNC, and year of birth, month of birth and gender from the NPD.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAge at first offence\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eIndividuals with missing or incorrect age in the PNC (age \u0026lt;10 years or age \u0026gt;31 years) were removed (N=6,144 [0.9%]). Age was then verified using year and month of birth in the NPD and individuals with inconsistent age records were removed (N=3,598 [0.5%]). Age at first offence was calculated by taking the earliest recorded offence age for each individual in the PNC. We then created a categorical age at first offence variable, with the following categories: Child (10-13 years), Adolescent (14-17 years), Young adult (18-20 years) and Adult (\u0026sup3;21 years). These age categories correspond with previous research on offending trajectories in a mixed-gender sample using this linked data \u003csup\u003e24\u003c/sup\u003e. These age categories also align with research that defines developmental age groups \u003csup\u003e47,48\u003c/sup\u003e and categories used by the MoJ to define prolific offenders \u003csup\u003e49\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAge at last offence\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAge at last offence was created by taking the latest recorded offence age for each individual. We then created a categorical age at last offence variable with the following age categories: Juvenile (10-17 years), Young adult (18-20 years) and Adult (\u0026sup3;21 years). These are the same age categories used by the MoJ to define prolific offenders \u003csup\u003e49\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eOffence history\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe created a series of persistent offender variables which indicate whether an individual had committed more than the median number of offences (excluding zero) as a juvenile, young adult and adult. The median total number of offences committed by females from age 10-17 years was one, therefore a persistent juvenile offender was one who had committed two or more offences as a juvenile. A persistent young adult female offender was defined as an individual who had committed three or more offences from age 18-20 years. A persistent adult female offender was defined as an individual who had committed five or more offences from age 21 years onwards.\u003c/p\u003e\n\u003cp\u003eThis allows an individual to be persistent in one age category but not persistent in another, or persistent in all age categories. This definition was used for a previous publication using this data\u003csup\u003e24\u003c/sup\u003e. It is distinct from the MoJ\u0026rsquo;s\u003csup\u003e49\u003c/sup\u003e definition of a prolific offender, in which the categories are mutually exclusive.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eViolent offending\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eA violent offender was classified as an individual who had ever committed an offence defined as \u0026lsquo;Violence against the person\u0026rsquo; or a contact sexual offence (according to HO offence codes).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel covariates\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCustodial sentence\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eUsing PNC Disposal codes, a binary variable was created to indicate whether an individual had ever received a custodial sentence (not including suspended custodial sentences). This was used as a covariate, as individuals are largely unable to offend during periods of incarceration. This may mimic desistance and affect latent class membership, particularly for persistent offenders who are more likely to receive custodial sentences \u003csup\u003e50\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLatent class analysis (LCA) was employed to model female offending trajectories. LCA is a type of finite mixture modelling used to identify homogenous groups within heterogenous populations \u003csup\u003e51\u003c/sup\u003e. It uses patterns of observed indicators to identify unobserved or latent groups. To prevent the model converging on local rather than global maxima, the number of random starts was set to 1000, and the number of final stage optimizations was set to 100. A maximum of 50 iterations were allowed in the initial stage.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe following steps were performed to select the model that best fit the data \u003csup\u003e52\u003c/sup\u003e. A one-class (k) model was created, followed by models with k+1 latent classes, until models no longer converged. Log-likelihood, Bayesian information criteria (BIC), sample size adjusted BIC (aBIC) and Akaike information criteria (AIC) were recorded to compare model fit with subsequent models. The potential best-fitting models were then compared on these indicators. Models with the lowest BIC, aBIC and AIC values were favoured. These indicators give a measure of how well the model fits the data based on the penalized log-likelihood. Additionally, parsimony, class sizes, entropy and interpretability of classes within the context of previous literature were considered \u003csup\u003e53\u003c/sup\u003e. Posterior class probabilities were estimated and used to assign individuals to optimal groups based on the highest probability. Average posterior probabilities (AvePP) and odds of correct classification [OCC] are reported for classes in the final model. Values of over 0.8 for AvePP and 5 for OCC are considered good \u003csup\u003e54,55\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe variables described above, age at first offence, age at last offence and persistence as a juvenile, young adult and adult were used as indicator variables for all latent class models. Models were developed with and without the violent offender indicator variable, to assess whether this impacted the model. When the optimum number of classes was identified, the custodial sentence variable was added to the model as a covariate \u003csup\u003e56\u003c/sup\u003e. A sensitivity analysis was performed, rerunning the final 4-class model, but including individuals whose age was not verified in the NPD (due to missing or incorrect data), in order to assess the impact of excluding these individuals on the final model.\u0026nbsp;\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eThe demographic characteristics and offence characteristics of the final sample of female offenders are presented in Tables 1 and 2. Most individuals had their first interaction with the CJS as juveniles (age 10-17 years). This was also the most common age period to be classified as a persistent offender. Most individuals had never received a custodial sentence (96.1%) or committed a violent or contact sexual offence (86.9%). The most common offence type was summary offences excluding motoring (such as common assault and battery, betting and gambling, and television license offences) (40.0%), followed by theft offences (25.5%) (Table 2). Just 3.5% (N=34,084) of all offences resulted in a custodial sentence, mainly theft (31.9%) and summary offences excluding motoring (23.6%).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1.\u003c/strong\u003e \u003cem\u003eOffending and demographic characteristics of sample (N=196,089)\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u003cem\u003eFirst offence:\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eChild (10-13 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e48,980 (25.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eAdolescent (14-17 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e91,699 (46.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eYoung adult (18-20 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e29,082 (14.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eAdult (21-31 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e26,328 (13.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 502px;\"\u003e\n \u003cp\u003e\u003cem\u003eLast offence:\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eJuvenile (10-17 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e106,198 (54.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eYoung adult (18-20 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e38,536 (19.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eAdult (21-31 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e51,355 (26.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 502px;\"\u003e\n \u003cp\u003e\u003cem\u003ePersistence:\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e2 or more offences as a Juvenile (10-17 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e60,686 (30.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e3 or more offences as a Young adult (18-20 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e25,265 (12.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e5 or more offences as an Adult (21-31 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e24,235 (12.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eEver been incarcerated\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e7,662 (3.9%) \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eEver committed a violent or contact sexual offence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e25,656 (13.1%) \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u003cem\u003eEthnicity:\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eWhite\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e167,656 (85.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eBlack\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e10,200 (5.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eAsian\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e5,609 (2.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eMixed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e9,282 (4.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eOther\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e1,493 (0.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e1,849 (0.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eEver eligible for FSM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e103,562 (52.8%) \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 350px;\"\u003e\n \u003cp\u003eMedian IDACI score\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 152px;\"\u003e\n \u003cp\u003e0.25\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNote:\u0026nbsp;\u003c/em\u003eFree school meals (FSM) (population baseline rates of eligibility are around 25% \u003csup\u003e57\u003c/sup\u003e); Income deprivation affecting children index (IDACI) \u0026ndash; indicates that, as children, individuals in the sample lived in areas where on average 25.0% of children aged 0-15 were living in income-deprived families (in 2019, the highest local authority IDACI score was 0.32 \u003csup\u003e44\u003c/sup\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2.\u003c/strong\u003e \u003cem\u003eCounts of offence type and offences resulting in a custodial sentence.\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"614\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003eCounts of offences (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003eCounts of offences resulting in custodial sentences (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e01 Violence against the person\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e64,051 (6.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e3,949 (11.6)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e02 Sexual offences\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e1,112 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e260 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e03 Robbery\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e10,501 (1.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e1,452 (4.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e04 Theft Offences \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e246,903 (25.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e10,881 (31.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e05 Criminal damage and arson \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e11,269 (1.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e403 (1.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e06 Drug offences \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e42,634 (4.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e1,916 (5.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e07 Possession of weapons \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e13,245 (1.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e1,025 (3.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e08 Public order offences \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e17,834 (1.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e1,326 (3.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e09 Miscellaneous crimes against society \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e37,430 (3.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e3,223 (9.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e10 Fraud offences \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e24,052 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e1,122 (3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e11 Summary offences excluding motoring \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e386,267 (40.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e8,060 (23.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e12 Summary motoring offences\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e109,872 (11.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e251 (0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e20 Unknown offences\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e1,311 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e216 (0.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 274px;\"\u003e\n \u003cp\u003e\u003cem\u003eTotal\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003e966,481 (100.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003e34,084 (100.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eWe estimated models from one to five classes. The addition of an extra class in the five-class model did not improve interpretability in the context of the previous literature, therefore results are not reported. The model also failed to converge without significantly increasing the number of random starts. \u0026nbsp;Goodness of fit statistics for the one- to four-class model are reported in Table 3.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3.\u003c/strong\u003e \u003cem\u003eModel fit statistics for 1 to 4 class model (n=196,089)\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003edf\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eLog-likelihood\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003eBIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003eaBIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eAIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003eEntropy\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003eClass sizes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003eModel 1\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003e87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e-712629.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e1425356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003e1425331\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1425274\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003eModel 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003e78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e-611005.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e1222218\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003e1222164\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1222044\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e46/54\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003eModel 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003e69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e-563364.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e1127045\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003e1126962\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1126780\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e54/20/26\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003eModel 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e-538191.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e1076810\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003e1076699\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e1076453\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e54/16/11/19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNote:\u003c/em\u003e df: degrees of freedom; BIC: Bayesian information criterion; aBIC: adjusted Bayesian information criterion; AIC: Akaike\u0026rsquo;s information criterion.\u003c/p\u003e\n\u003cp\u003eBased on having the lowest BIC, AIC and aBIC values, class sizes and the interpretability of classes, a four-class model was selected as the best-fitting model. Adding the \u003cem\u003eViolent offending\u003c/em\u003e variable to the model did not affect the number or nature of classes identified, and it was therefore not included in the final model. The classes\u0026rsquo; interpretation and sizes remained similar after including \u003cem\u003ecustodial sentence\u003c/em\u003e as a covariate. Fit statistics for the final four-class model, controlling for having ever served a custodial sentence, are reported in Table 4. All average posterior probabilities were over 0.9.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe profiles of the identified trajectories after adjusting for having served a custodial sentence can be seen below in Table 5. The \u0026lsquo;Life-course-persistent\u0026rsquo; (\u0026lsquo;LCP\u0026rsquo;) group was the smallest group identified, accounting for 11% of the sample. These individuals began offending across childhood and adolescence and continued offending into adulthood. These individuals were persistent as juveniles and showed mixed levels of persistence in young adulthood and adulthood. \u0026lsquo;Adolescent-limited\u0026rsquo; offenders were the largest group, accounting for 54% of the sample. These individuals began offending as juveniles (mainly in adolescence) and had desisted when they reached adulthood. Persistence as juveniles was mixed. The \u0026lsquo;Young-adult-limited\u0026apos; group accounted for 16% of the sample. These individuals offended in young adulthood only and persistence was mixed during this period. \u0026lsquo;Adult-onset\u0026rsquo; offenders accounted for 19% of the sample. These individuals began offending after the age of 21 and were mixed in terms of persistence during adulthood.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4.\u003c/strong\u003e \u003cem\u003eFinal model classes\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 194px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e\u003cstrong\u003emcap\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAvePP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOCC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 194px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClass\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 194px;\"\u003e\n \u003cp\u003eLife-course-persistent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e0.11; 22,118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83px;\"\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 92px;\"\u003e\n \u003cp\u003e144.57\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 194px;\"\u003e\n \u003cp\u003eAdolescent-limited\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e0.54; 106,198\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 92px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 194px;\"\u003e\n \u003cp\u003eYoung-adult-limited\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e0.16; 31,055\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83px;\"\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 92px;\"\u003e\n \u003cp\u003e106.45\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 194px;\"\u003e\n \u003cp\u003eAdult-onset\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e0.19; 36,718\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83px;\"\u003e\n \u003cp\u003e0.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 92px;\"\u003e\n \u003cp\u003e110.94\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNote:\u003c/em\u003e mcap: max probability class assignment proportion; AvePP: Average posterior probability, OCC: odds of correct classification\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 5.\u003c/strong\u003e \u003cem\u003eResponse parameter probabilities for 4 latent classes\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"622\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTrajectory:\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLife-course-persistent\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e11% (N=22,118)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAdolescent-limited\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e54% (N=106,198)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eYoung-adult-limited\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e16% (N=31,055)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAdult-onset\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e19% (N=36,718)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 622px;\"\u003e\n \u003cp\u003e\u003cem\u003eIndicator variables:\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eFirst offence childhood (10-13 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.03\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.04\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eFirst offence adolescence (14-17 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.12\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.14\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eFirst offence young adulthood (18-20 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.85\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.10\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eFirst offence adulthood (\u0026gt;21 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.74\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eLast offence juvenile (10-17 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.98\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eLast offence young adult (18-20 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.02\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.96\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eLast offence adulthood (\u0026gt;21 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.01\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.04\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eJuvenile persistence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.95\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eYoung adult persistence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 178px;\"\u003e\n \u003cp\u003eAdult persistence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.00\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNote:\u0026nbsp;\u003c/em\u003e%s below 0.30 or above 0.70 are presented in bold\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSensitivity analysis\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe sensitivity analysis, including individuals whose ages could not be verified using NPD data, confirmed that excluding individuals with inconsistent ages in the NPD did not affect results (see Supplementary Table S2 for details of the sensitivity analysis model).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCharacteristics of offending trajectories\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTable 6 shows the offending characteristics of the sample according to trajectory membership. \u0026apos;LCP\u0026rsquo; offenders were responsible for the largest proportion of offences (42%) despite making up the smallest proportion of the sample. Additionally, 19% of \u0026lsquo;LCP\u0026rsquo; offenders had received a custodial sentence, compared to less than 1% of \u0026lsquo;Adolescent-limited\u0026rsquo; offenders. Three percent of \u0026lsquo;Young-adult limited\u0026rsquo; and five percent of \u0026lsquo;Adult-onset\u0026rsquo; offenders, had received a custodial sentence. On average, \u0026apos;LCP\u0026rsquo; offenders had records for 11 offences across the study period, while \u0026lsquo;Adult-onset\u0026rsquo; offenders had four, \u0026lsquo;Young-adult limited\u0026rsquo; had two and \u0026lsquo;Adolescent-limited\u0026rsquo; had one. For a breakdown of total offences by offence type and trajectory type, see Supplementary Table S3.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 6.\u003c/strong\u003e \u003cem\u003eOffence and education characteristics according to latent classes\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"628\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eTrajectory:\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLife-course-persistent\u0026nbsp;\u003c/strong\u003e(N=22,118)\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAdolescent-limited\u003c/strong\u003e (N=106,198)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eYoung-Adult-limited\u003c/strong\u003e (N=31,055)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAdult-onset\u003c/strong\u003e (N=36,718)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003e\u003cem\u003eTotal offences committed\u003c/em\u003e (% of cohort offences)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e403,867 (41.8)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e237,823 (24.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e108,555 (11.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e218,836 (22.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eEver incarcerated\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e4,237 (19.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e667 (0.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e933 (3.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,825 (5.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eReceived a custodial sentence under 12 months\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e3,282 (14.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e445 (0.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e506 (1.6%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,099 (3.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eSerious violent offenders\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e7,780 (35.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e11,348 (10.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e2,424 (7.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e3,006 (8.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eMedian total offences per person (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e11 (16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1 (1)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e2 (4)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e4 (5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eMean age first offence (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e13.8 (1.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e14.1 (1.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e18.1 (1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e21.3 (3.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eMean age last offence (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e22.2 (3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e14.6 (1.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e19.1 (1.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e23.7 (2.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eMedian criminal career duration (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e8 (6)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1 (0)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1 (0)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1 (3)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eEver been eligible for free school meals (FSM)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e14,936 (67.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e55,733 (52.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e15,195 (48.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e17,698 (48.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eMedian Income Deprivation Affecting Children (IDACI) score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e0.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e0.25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e0.24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e0.24\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 628px;\"\u003e\n \u003cp\u003e\u003cem\u003eEthnicity:*\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eWhite\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e19,049 (86.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e92,396 (87.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e25,779 (83%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e30,432 (82.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eBlack\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,105 (5.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e4,624 (4.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e2,125 (6.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e2,346 (6.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eAsian\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e274 (1.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e2,821 (2.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,107 (3.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,407 (3.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eMixed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,356 (6.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e4,710 (4.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,458 (4.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e1,758 (4.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 163px;\"\u003e\n \u003cp\u003eOther\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e121 (0.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e728 (0.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e277 (0.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 116px;\"\u003e\n \u003cp\u003e367 (1.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e*Counts do not add up to total N of latent classes due to missing data (all classes had around 1% missing ethnicity data). Percentages may not add up to 100% due to rounding.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe median IDACI score in the \u0026apos;LCP\u0026rsquo; group was 0.29; meaning that, as children, these female offenders lived in neighbourhoods where on average 29% of children lived in families with low-income. The other groups had similar median IDACI scores ranging from 0.24-0.25. Additionally, 68% of individuals in the \u0026apos;LCP\u0026apos; trajectory had received free school meals at some point whilst at school, compared to 48-52% of individuals in the other trajectories. See Table Supplementary S4 for further demographic and offence characteristics of latent classes.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eFemales have long been neglected in research on life course patterns of offending. It is key that we look at these patterns as they may allow us to distinguish between groups of offenders with distinct risk and recidivism factors and treatment/disposal needs. We used a large administrative sample from linked PNC and NPD data to identify subgroups of female offenders from the age of criminal responsibility (age 10 years) up to age 31 years. This is the largest study to date specifically exploring female offending trajectories and the first in England to explore female-only trajectories of reoffending from childhood well into adulthood with a nationally representative sample. We used latent class analysis to identify four groups with different patterns of onset, desistance and rates of offending: Life-course-persistent, Adolescent-limited, Young-Adult-limited and Adult-onset. While these groups resemble those found in previous literature using both male and female samples \u003csup\u003e21,58\u003c/sup\u003e, the characteristics and associated risk factors for the identified groups are likely to differ between males and females. It is noteworthy that over 50% of the offending in this cohort was for summary offences (including motoring) and that these summary offences and theft accounted for over 76% of the offences, but that considerable serious offending was still observed. The findings contribute to our understanding of heterogeneity in female offending and demonstrate the need for developmentally and trajectory appropriate interventions.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe identified an LCP trajectory, resembling Moffitt\u0026rsquo;s\u003csup\u003e17\u003c/sup\u003e LCP group, that accounted for 11% of the sample. These offenders made up the majority of the incarcerated population in the sample (of the 7,662 individuals who had ever been incarcerated, 55% were classified as LCP offenders). This may reflect the high average number of offences committed by this group (11 offences per offender), as previous convictions are considered in sentencing decisions \u003csup\u003e59\u003c/sup\u003e. Fifteen percent of LCP offenders had received custodial sentences of under 12 months. Such sanctions are commonly given to female offenders and are associated with increased reoffending rates compared to community-based sentences \u003csup\u003e6,60\u003c/sup\u003e, although the baseline characteristics of these two groups may differ \u003csup\u003e15\u003c/sup\u003e. These offenders also displayed high rates of socioeconomic disadvantage with 68% eligible for FSMs compared to around 25% in the general population \u003csup\u003e57\u003c/sup\u003e. As this group accounts for the largest share of offences among female offenders, they should be a priority for interventions. Early intensive interventions focusing on family \u003csup\u003e61\u003c/sup\u003e and school environments \u003csup\u003e62\u003c/sup\u003e may be beneficial for this group.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFemale persistent offending is more likely to be linked to the antisocial influence of partners, mental health disorders and exposure to violence compared to male persistence \u003csup\u003e26\u003c/sup\u003e. Qualitative perspectives also highlight the role of victimisation in driving young girls to commit particular offences (e.g. petty property offences and offences related to prostitution) \u003csup\u003e63\u003c/sup\u003e. Recent research modeling offending trajectories among care-experienced individuals found that females who experienced out of home placements during adolescence were at the highest risk of becoming persistent offenders \u003csup\u003e64,65\u003c/sup\u003e. Future research should further explore the early correlates and predictors of membership in this group. This may help to prospectively identify at-risk females for interventions in schools or other education settings, prior to offending onset. Interventions should target the gender-specific needs of high-risk females (e.g. family separation, substance misuse, and disconnection from school \u003csup\u003e66\u003c/sup\u003e). This could allow us to more effectively distribute resources to offenders at higher risk of negative life outcomes. This may help to prevent the onset of these high-rate criminal careers, reducing the associated individual, societal, and inter-generational costs.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe Adolescent-limited offending group identified only committed offences as juveniles and made up around 54% of the total sample. This indicates that, in line with Moffitt\u0026rsquo;s theory, this is the most common pattern of offending amongst females \u003csup\u003e67\u003c/sup\u003e. It is unclear why these offenders desist, but this may be due to a narrowing of the gap between social and biological maturity over time \u003csup\u003e17\u003c/sup\u003e. Qualitative work exploring this adolescent group has identified delinquent peers and identity exploration as important factors for offending in this group \u003csup\u003e68\u003c/sup\u003e. While this group appear to have significantly better later life outcomes than persistent groups \u003csup\u003e69\u003c/sup\u003e, they do show deficits in other areas, such as educational attainment \u003csup\u003e70\u003c/sup\u003e, indicating a need for support.\u003c/p\u003e\n\u003cp\u003eThe Young-adult-limited group identified in this study made up 16% of the total sample. This group has recently been identified in male offender samples \u003csup\u003e64,65\u003c/sup\u003e. They had higher rates of incarceration (3%) than Adolescent-limited offenders (0.6%). This may be due to differences in sentencing guidelines between adults and minors. Similar to the Adolescent-limited group, these offenders appeared to age out of offending as they matured. This group may have more protective factors than their persistent counterparts, e.g. active employment \u003csup\u003e71\u003c/sup\u003e. Previous research on female offenders has found that this group had similar risk factors to very low-level sporadic offenders \u003csup\u003e71\u003c/sup\u003e. They also experienced different outcomes later in life; for example, the majority of late-onset desisting offenders who married later experienced a divorce. Late adolescence may be a critical window for interventions in this group. Due to their age, it is key that this group does not fall in the gap between the rehabilitation-oriented youth justice system and the more punitive adult justice system Services that support women in this transitional age period (e.g. with employment, housing, vocational skills) may help to prevent the onset of offending for these individuals.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAdult-onset groups have caused disagreements in previous literature, with some research maintaining that this group is an artifact of methodological factors and unidentified juvenile offences \u003csup\u003e18\u003c/sup\u003e. In our sample this group made up 19% of offenders. The late onset of this group indicates the need to investigate adulthood risk factors for offending onset in females, and to potentially target specific interventions later in adolescence or early adulthood. \u0026nbsp;These offenders may represent \u0026lsquo;late-bloomers\u0026rsquo; \u003csup\u003e72\u003c/sup\u003e, a group who start offending during adulthood but reach the same level of offending as LCP offenders in later life. However, previous research also finds adult-onset groups that desist over time \u003csup\u003e71\u003c/sup\u003e. As our sample was only followed-up until age 31 years, it is unclear whether this group persists or desists throughout later adulthood. The later-onset of this group (at or after 21 years) may be explained by the presence of coping strategies and support networks in adolescence and early adulthood, that stop being effective as they experience the reduced social support and increasing demands that come with later adulthood \u003csup\u003e73\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eLimitations\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFinite mixture modelling (FMM) approaches, including latent class analysis, have typically been used to examine longitudinal patterns of offending \u003csup\u003e19\u003c/sup\u003e. Criticisms of FMM include the potential misinterpretation of identified groups as \u0026lsquo;real\u0026rsquo; subgroups \u003csup\u003e74\u003c/sup\u003e. It should be acknowledged that groups identified using FMM are not real entities but probabilistic approximations of patterns in outcomes and should therefore be interpreted accordingly.\u003c/p\u003e\n\u003cp\u003eAdministrative offending records only account for offences that have been dealt with by the CJS. Offending rates are likely to be underestimated, particularly for young female offenders, who may be less likely to receive official sanctions due to committing less serious offences \u003csup\u003e75\u003c/sup\u003e. It is estimated that the actual age of offending onset is around 3-5 years prior to an individual\u0026rsquo;s first recorded offence \u003csup\u003e76\u003c/sup\u003e. Official measures may also be a proxy for more serious offences as \u0026lsquo;normative\u0026rsquo; adolescent offending may not be captured in these records \u003csup\u003e72\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe maximum follow-up in this study was to age 31 years. While this is an improvement on much previous research (most studies that continue into adulthood do not follow-up past age 26), ideally follow-up would be further into adulthood. As the birth cohort used in this research ages, this information will become available for further analysis. Additionally, we have an uneven length of follow-up for individuals due to the seven-year span of the birth cohort (1990-1997). This means that our study end-point ranged from 24 to 31 years of age. This has the potential to influence latent class assignment: however, individuals in each birth year are relatively evenly distributed between identified trajectories (see Supplementary Table S4).\u003c/p\u003e\n\u003cp\u003eWe could not control for mortality or migration, as neither the NPD nor PNC contain information on this. This may impact the groups identified, because if someone dies or emigrates, they will appear to have desisted from offending. Premature aging and early death are more common amongst persistent offenders \u003csup\u003e77,78\u003c/sup\u003e, therefore this may have resulted in an under-identification of LCP offenders.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFuture Directions\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe current study exemplifies the power of using administrative data to answer questions related to public policy, using near whole-population samples with relatively low financial and time burdens. Recent work using UK administrative data, for example looking at changes in educational attainment and criminal offending \u003csup\u003e79\u003c/sup\u003e and the intersection between care experience, ethnicity and offending \u003csup\u003e80\u003c/sup\u003e, shows the potential of using linked datasets to explore social justice issues. Further research on the social care and educational backgrounds of female offenders, particularly the LCP group identified, would help to guide research focused on gender-informed approaches that incorporate a Whole Systems Approach model from childhood onwards \u003csup\u003e81\u003c/sup\u003e.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study has enhanced our theoretical understanding of female offending trajectories through harnessing a large administrative sample of female offenders convicted or cautioned for an offence in England between 2000 and 2020. This is the largest study looking at female-only trajectories to date. By using a large nationally-representative dataset we move beyond small-scale studies and provide a foundation for further gender-informed criminal justice policy research. \u0026nbsp;We identified four distinct trajectories of offending among females and explored offence and some demographic characteristics of these groups. Our findings highlight the need for gender-responsive strategies that account for distinct female offending trajectories. Through developing our theoretical understanding of female offending patterns, we can inform criminal justice interventions to support and rehabilitate female offenders.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFunding\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCC received support for this work from Administrative Data Research (ADR) UK, an Economic and Social Research Council (ESRC) investment (Grant number: ES/P000703/1).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAcknowledgements\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was undertaken in the Office for National Statistics Secure Research Service using data from the ONS and other owners and does not imply the endorsement of the ONS or other data owners.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAuthor contributions\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthors HD, AW and CC contributed to the article conceptualisation. CC performed the data cleaning and analysis and prepared the original draft. CC was the only author to have access to the dataset. Authors AW, DJ, NB, and HD critically reviewed and edited the work. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData availability statement\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData used for this research is accessible via an application by an ONS accredited researcher to the data owners (Ministry of Justice and Department for Education).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAdditional Information (including a Competing Interests Statement)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAW and NB are supported by the National Institute for Health and Care Research (NIHR) Maudsley Biomedical Research Centre (BRC). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eFair, H. \u0026amp; Walmsley, R. World Female Imprisonment List. (Institute for Crime and Justice Policy Research (ICPR), London, UK, 2022).\u003c/li\u003e\n\u003cli\u003eFavril, L., Rich, J. D., Hard, J. \u0026amp; Fazel, S. Mental and physical health morbidity among people in prisons: an umbrella review. \u003cem\u003eThe Lancet. Public health\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, e250-e260 (2024). https://doi.org/10.1016/S2468-2667(24)00023-9\u003c/li\u003e\n\u003cli\u003eMcLeod, K. E.\u003cem\u003e et al.\u003c/em\u003e Health conditions among women in prisons: a systematic review. \u003cem\u003eThe Lancet. Public health\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, e609-e624 (2025). https://doi.org/10.1016/S2468-2667(25)00092-1\u003c/li\u003e\n\u003cli\u003eNilsson, S. F., Nordentoft, M., Fazel, S. \u0026amp; Laursen, T. M. Risk of homelessness after prison release and recidivism in Denmark: a nationwide, register-based cohort study. \u003cem\u003eThe Lancet. Public health\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, e756-e765 (2023). https://doi.org/10.1016/S2468-2667(23)00152-4\u003c/li\u003e\n\u003cli\u003eKarlsson, M. E. \u0026amp; Zielinski, M. J. Sexual Victimization and Mental Illness Prevalence Rates Among Incarcerated Women: A Literature Review. \u003cem\u003eTrauma, violence \u0026amp; abuse\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 326-349 (2020). https://doi.org/10.1177/1524838018767933\u003c/li\u003e\n\u003cli\u003eMinistry of Justice. Female Offender Strategy. (HMSO, London, 2018).\u003c/li\u003e\n\u003cli\u003eKelman, J., Gribble, R., Harvey, J., Palmer, L. \u0026amp; MacManus, D. How Does a History of Trauma Affect the Experience of Imprisonment for Individuals in Women\u0026apos;s Prisons: A Qualitative Exploration. \u003cem\u003eWomen \u0026amp; criminal justice\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, 171-191 (2024). https://doi.org/10.1080/08974454.2022.2071376\u003c/li\u003e\n\u003cli\u003eMinistry of Justice. Safety in custody quarterly: update to December 2017. (2018).\u003c/li\u003e\n\u003cli\u003ePrison Reform Trust. \u003cem\u003eResetting the approach to women\u0026rsquo;s imprisonment\u003c/em\u003e, \u0026lt;https://prisonreformtrust.org.uk/wp-content/uploads/2025/04/Resetting-the-approach-2025-briefing.pdf\u0026gt; (2025).\u003c/li\u003e\n\u003cli\u003eLee, R. D., Fang, X. \u0026amp; Luo, F. The impact of parental incarceration on the physical and mental health of young adults. \u003cem\u003ePediatrics (Evanston)\u003c/em\u003e \u003cstrong\u003e131\u003c/strong\u003e, e1188-e1195 (2013). https://doi.org/https://doi.org/10.1542/peds.2012-0627\u003c/li\u003e\n\u003cli\u003eMurray, J., Farrington, D. P. \u0026amp; Sekol, I. Children\u0026apos;s Antisocial Behavior, Mental Health, Drug Use, and Educational Performance After Parental Incarceration: A Systematic Review and Meta-Analysis. \u003cem\u003ePsychological bulletin\u003c/em\u003e \u003cstrong\u003e138\u003c/strong\u003e, 175-210 (2012). https://doi.org/https://doi.org/10.1037/a0026407\u003c/li\u003e\n\u003cli\u003ePoehlmann, J. Representations of Attachment Relationships in Children of Incarcerated Mothers. \u003cem\u003eChild development\u003c/em\u003e \u003cstrong\u003e76\u003c/strong\u003e, 679-696 (2005). https://doi.org/10.1111/j.1467-8624.2005.00871.x\u003c/li\u003e\n\u003cli\u003eGelsthorpe, L. in \u003cem\u003eWomen who offend\u003c/em\u003e (ed Gill McIvor) (Jessica Kingsley Publishers, 2004).\u003c/li\u003e\n\u003cli\u003eSteffensmeier, D. \u0026amp; Allan, E. Gender and Crime: Toward a Gendered Theory of Female Offending. \u003cem\u003eAnnual review of sociology\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 459-487 (1996). https://doi.org/10.1146/annurev.soc.22.1.459\u003c/li\u003e\n\u003cli\u003eYukhnenko, D., Wolf, A., Blackwood, N. \u0026amp; Fazel, S. Recidivism rates in individuals receiving community sentences: A systematic review. \u003cem\u003ePloS one\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, e0222495 (2019). https://doi.org/10.1371/journal.pone.0222495\u003c/li\u003e\n\u003cli\u003ePiquero, A. R. \u0026ldquo;We study the past to understand the present; we understand the present to guide the future\u0026rdquo;: The time capsule of developmental and life-course criminology. \u003cem\u003eJournal of criminal justice\u003c/em\u003e \u003cstrong\u003e85\u003c/strong\u003e, 101932 (2023). https://doi.org/10.1016/j.jcrimjus.2022.101932\u003c/li\u003e\n\u003cli\u003eMoffitt, T. E. Adolescence-Limited and Life-Course-Persistent Antisocial Behavior: A Developmental Taxonomy. \u003cem\u003ePsychological review\u003c/em\u003e \u003cstrong\u003e100\u003c/strong\u003e, 674-701 (1993). https://doi.org/https://doi.org/10.1037/0033-295X.100.4.674\u003c/li\u003e\n\u003cli\u003eMoffitt, T. E. Male antisocial behaviour in adolescence and beyond. \u003cem\u003eNature human behaviour\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 177-186 (2018). https://doi.org/https://doi.org/10.1038/s41562-018-0309-4\u003c/li\u003e\n\u003cli\u003eNagin, D. S. \u0026amp; Land, K. C. AGE, CRIMINAL CAREERS, AND POPULATION HETEROGENEITY: SPECIFICATION AND ESTIMATION OF A NONPARAMETRIC, MIXED POISSON MODEL. \u003cem\u003eCriminology (Beverly Hills)\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 327-362 (1993). https://doi.org/https://doi.org/10.1111/j.1745-9125.1993.tb01133.x\u003c/li\u003e\n\u003cli\u003eSilverthorn, P., Frick, P. J. \u0026amp; Reynolds, R. Timing of onset and correlates of severe conduct problems in adjudicated girls and boys. \u003cem\u003eJournal of Psychopathology and Behavioral Assessment\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 171-181 (2001). https://doi.org/https://doi.org/10.1023/A:1010917304587\u003c/li\u003e\n\u003cli\u003eJennings, W. G. \u0026amp; Reingle, J. M. On the number and shape of developmental/life-course violence, aggression, and delinquency trajectories: A state-of-the-art review. \u003cem\u003eJournal of criminal justice\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 472-489 (2012). https://doi.org/https://doi.org/10.1016/j.jcrimjus.2012.07.001\u003c/li\u003e\n\u003cli\u003eJolliffe, D., Farrington, D. P., Piquero, A. R., MacLeod, J. F. \u0026amp; van de Weijer, S. Prevalence of life-course-persistent, adolescence-limited, and late-onset offenders: A systematic review of prospective longitudinal studies. \u003cem\u003eAggression and violent behavior\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 4-14 (2017). https://doi.org/10.1016/j.avb.2017.01.002\u003c/li\u003e\n\u003cli\u003eJolliffe, D., Farrington, D. P., Piquero, A. R., Loeber, R. \u0026amp; Hill, K. G. Systematic review of early risk factors for life-course-persistent, adolescence-limited, and late-onset offenders in prospective longitudinal studies. \u003cem\u003eAggression and violent behavior\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 15-23 (2017). https://doi.org/10.1016/j.avb.2017.01.009\u003c/li\u003e\n\u003cli\u003eDickson, H. \u003cem\u003eChildhood educational predictors of reoffending trajectories\u003c/em\u003e, \u0026lt;https://www.adruk.org/fileadmin/uploads/adruk/Documents/Data_Insights/Data-Insight-Childhood-educational-predictors-re-offending-trajectories.pdf\u0026gt; (2024).\u003c/li\u003e\n\u003cli\u003eGushue, K., McCuish, E. C. \u0026amp; Corrado, R. R. Developmental Offending Patterns: Female Offending Beyond the Reference Category. \u003cem\u003eCriminal justice and behavior\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, 139-156 (2021). https://doi.org/10.1177/0093854820966729\u003c/li\u003e\n\u003cli\u003eCauffman, E., Monahan, K. C. \u0026amp; Thomas April, G. Pathways to Persistence: Female Offending from 14 to 25. \u003cem\u003eJournal of Developmental and Life-Course Criminology\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 236-268 (2015). https://doi.org/https://doi.org/10.1007/s40865-015-0016-z\u003c/li\u003e\n\u003cli\u003eOdgers, C. L.\u003cem\u003e et al.\u003c/em\u003e Female and male antisocial trajectories: From childhood origins to adult outcomes. \u003cem\u003eDevelopment and psychopathology\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 673-716 (2008). https://doi.org/https://doi.org/10.1017/S0954579408000333\u003c/li\u003e\n\u003cli\u003eSivertsson, F., Carlsson, C., Almquist, Y. B. \u0026amp; Br\u0026auml;nnstr\u0026ouml;m, L. Offending trajectories from childhood to retirement age: Findings from the Stockholm birth cohort study. \u003cem\u003eJournal of criminal justice\u003c/em\u003e \u003cstrong\u003e91\u003c/strong\u003e, 102155 (2024). https://doi.org/https://doi.org/10.1016/j.jcrimjus.2024.102155\u003c/li\u003e\n\u003cli\u003eAndersson, F., Levander, S., Svensson, R. \u0026amp; Levander, M. T. Sex differences in offending trajectories in a Swedish cohort: Sex differences in offending trajectories. \u003cem\u003eCriminal behaviour and mental health\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 108-121 (2012). https://doi.org/https://doi.org/10.1002/cbm.1822\u003c/li\u003e\n\u003cli\u003eCarlsson, C. \u0026amp; Sivertsson, F. Age, gender, and crime in a Stockholm birth cohort to age 64. \u003cem\u003eJournal of Developmental and Life-Course Criminology\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 359-384 (2021). \u003c/li\u003e\n\u003cli\u003eFergusson, D. M. \u0026amp; Horwood, L. J. Male and female offending trajectories. \u003cem\u003eDevelopment and psychopathology\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 159-177 (2002). https://doi.org/https://doi.org/10.1017/S0954579402001098\u003c/li\u003e\n\u003cli\u003eKuluk, A., Allard, T., Thompson, C., Ogilvie, J. M. \u0026amp; Broidy, L. Offending Trajectories in an Australian Birth Cohort: Differences and Similarities Across Sex. \u003cem\u003eCriminal justice and behavior\u003c/em\u003e \u003cstrong\u003e51\u003c/strong\u003e, 807-830 (2024). https://doi.org/10.1177/00938548241234373\u003c/li\u003e\n\u003cli\u003eCauffman, E. Understanding the Female Offender. \u003cem\u003eThe Future of children\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 119-142 (2008). https://doi.org/10.1353/foc.0.0015\u003c/li\u003e\n\u003cli\u003eMinistry of Justice. Statistics on Women and the Criminal Justice System 2023. (2025).\u003c/li\u003e\n\u003cli\u003eBenchimol, E. I.\u003cem\u003e et al.\u003c/em\u003e The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD). \u003cem\u003ePLoS Med\u003c/em\u003e (2015). \u003c/li\u003e\n\u003cli\u003evon Elm, E.\u003cem\u003e et al.\u003c/em\u003e The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. \u003cem\u003eEpidemiology (Cambridge, Mass.)\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 800-804 (2007). https://doi.org/10.1097/EDE.0b013e3181577654\u003c/li\u003e\n\u003cli\u003eMplus User\u0026rsquo;s Guide (Muth\u0026eacute;n \u0026amp; Muth\u0026eacute;n, Los Angeles, CA, 1998-2017).\u003c/li\u003e\n\u003cli\u003eR: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, Austria, 2024).\u003c/li\u003e\n\u003cli\u003eRStudio: Integrated Development Environment for R (Posit Software, PBC, Boston, MA, 2024).\u003c/li\u003e\n\u003cli\u003eWickersham, A.\u003cem\u003e et al.\u003c/em\u003e Educational attainment trajectories among children and adolescents with depression, and the role of sociodemographic characteristics: longitudinal data-linkage study. \u003cem\u003eBritish journal of psychiatry\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e, 151-157 (2021). https://doi.org/https://doi.org/10.1192/bjp.2020.160\u003c/li\u003e\n\u003cli\u003eMinistry of Justice. A Technical Guide to Statistics on Race and the Criminal Justice System. (2018).\u003c/li\u003e\n\u003cli\u003eDepartment for Education. Education, children\u0026rsquo;s social care and offending: multi-level modelling: technical report., (2023).\u003c/li\u003e\n\u003cli\u003eDepartment for Education and Ministry of Justice. Education, children\u0026rsquo;s social care and offending. (2022).\u003c/li\u003e\n\u003cli\u003eMinistry of Housing, C. L. G. The English Indices of Deprivation 2019 (IoD2019). (2019).\u003c/li\u003e\n\u003cli\u003eCrawford, C. \u0026amp; Greaves, E. A comparison of commonly used socio-economic indicators: their relationship to educational disadvantage and relevance to Teach First. (Institute for Fiscal Studies, 2013).\u003c/li\u003e\n\u003cli\u003ePrison Reform Trust. \u003cem\u003eWhy focus on reducing women\u0026rsquo;s imprisonment?\u003c/em\u003e, \u0026lt;https://prisonreformtrust.org.uk/publication/why-focus-on-reducing-womens-imprisonment-3/\u0026gt; (2022).\u003c/li\u003e\n\u003cli\u003eBodker, A. \u0026amp; Harris, D. A. Help-Seeking Behaviors of Young People with Problematic Sexualized Behaviors. \u003cem\u003eJournal of child sexual abuse\u003c/em\u003e, 1-20 (2024). https://doi.org/10.1080/10538712.2024.2415550\u003c/li\u003e\n\u003cli\u003eSampson, R. J. \u0026amp; Laub, J. H. LIFE-COURSE DESISTERS? TRAJECTORIES OF CRIME AMONG DELINQUENT BOYS FOLLOWED TO AGE 70*. \u003cem\u003eCriminology\u003c/em\u003e \u003cstrong\u003e41\u003c/strong\u003e, 555 (2003). https://doi.org/10.1111/j.1745-9125.2003.tb00997.x\u003c/li\u003e\n\u003cli\u003eMinistry of Justice. Prolific Offenders - Defining a Prolific Offender. (2017).\u003c/li\u003e\n\u003cli\u003eFarrington, D. P. \u0026amp; West, D. J. Criminal, penal and life histories of chronic offenders: risk and protective factors and early identification. \u003cem\u003eCriminal behaviour and mental health\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 492-523 (1993). https://doi.org/10.1002/cbm.1993.3.4.492\u003c/li\u003e\n\u003cli\u003eHerle, M.\u003cem\u003e et al.\u003c/em\u003e Identifying typical trajectories in longitudinal data: modelling strategies and interpretations. \u003cem\u003eEuropean journal of epidemiology\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 205-222 (2020). https://doi.org/10.1007/s10654-020-00615-6\u003c/li\u003e\n\u003cli\u003eMasyn, K. E. in \u003cem\u003eThe Oxford Handbook of Quantitative Methods \u003c/em\u003e(ed T. D. Little) (Oxford University Press, 2013).\u003c/li\u003e\n\u003cli\u003eNagin, D. \u003cem\u003eGroup-based modeling of development\u003c/em\u003e. 1st edn, (Harvard University Press, 2005).\u003c/li\u003e\n\u003cli\u003eNagin, D. S. \u0026amp; Odgers, C. L. Group-Based Trajectory Modeling (Nearly) Two Decades Later. \u003cem\u003eJournal of Quantitative Criminology\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 445 (2010). https://doi.org/https://doi.org/10.1007/s10940-010-9113-7\u003c/li\u003e\n\u003cli\u003eClark, S. \u0026amp; Muth\u0026eacute;n, B. Relating Latent Class Analysis Results to Variables not Included in the Analysis. (2009). \u003c/li\u003e\n\u003cli\u003eVermunt, J. K. Latent Class Modeling with Covariates: Two Improved Three-Step Approaches. \u003cem\u003ePolitical analysis\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 450-469 (2010). https://doi.org/10.1093/pan/mpq025\u003c/li\u003e\n\u003cli\u003eDepartment for Education. Schools, pupils and their characteristics. (2024).\u003c/li\u003e\n\u003cli\u003eFontaine, N., Carbonneau, R., Vitaro, F., Barker, E. D. \u0026amp; Tremblay, R. E. Research review: A critical review of studies on the developmental trajectories of antisocial behavior in females. \u003cem\u003eJournal of child psychology and psychiatry\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, 363-385 (2009). https://doi.org/10.1111/j.1469-7610.2008.01949.x\u003c/li\u003e\n\u003cli\u003eSentencing Council Guidelines. Imposition of community and custodial sentences - Effective from 1 April 2025. (2025).\u003c/li\u003e\n\u003cli\u003ePrison Reform Trust. \u003cem\u003eSix in 10 women sent to prison serve sentences of less than six months\u003c/em\u003e, \u0026lt;https://prisonreformtrust.org.uk/six-in-10-women-sent-to-prison-serve-sentences-of-less-than-six-months/\u0026gt; (2023).\u003c/li\u003e\n\u003cli\u003eLeijten, P.\u003cem\u003e et al.\u003c/em\u003e Research Review: Harnessing the power of individual participant data in a meta‐analysis of the benefits and harms of the Incredible Years parenting program. \u003cem\u003eJournal of child psychology and psychiatry\u003c/em\u003e \u003cstrong\u003e59\u003c/strong\u003e, 99-109 (2018). https://doi.org/10.1111/jcpp.12781\u003c/li\u003e\n\u003cli\u003eVeenman, B., Luman, M. \u0026amp; Oosterlaan, J. Efficacy of behavioral classroom programs in primary school. A meta-analysis focusing on randomized controlled trials. \u003cem\u003ePloS one\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, e0201779 (2018). https://doi.org/10.1371/journal.pone.0201779\u003c/li\u003e\n\u003cli\u003eChesney-Lind, M. \u0026amp; Shelden, R. G. \u003cem\u003eGirls, delinquency, and juvenile justice\u003c/em\u003e. Fourth edition. edn, (Wiley, 2014).\u003c/li\u003e\n\u003cli\u003eBr\u0026auml;nnstr\u0026ouml;m, L., Andershed, A.-K., Vinnerljung, B., Hjern, A. \u0026amp; Almquist, Y. B. Gender-specific trajectories of offending from adolescence until age 40 among individuals with experience of out-of-home care: A national cohort study. \u003cem\u003eDevelopmental child welfare\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 3-22 (2024). https://doi.org/10.1177/25161032231217265\u003c/li\u003e\n\u003cli\u003eHossein, G. A prospective examination of offending trajectories in men and women with out-of-home care experience followed up until retirement age. \u003cem\u003eChild Abuse \u0026amp; Neglect\u003c/em\u003e \u003cstrong\u003e161\u003c/strong\u003e, 107253 (2025). https://doi.org/https://doi.org/10.1016/j.chiabu.2025.107253\u003c/li\u003e\n\u003cli\u003eShepherd, S. M., Newton, D., Harries, C., Fix, R. L. \u0026amp; Fullam, R. An analysis of high-risk offending pathways for young females in custody. \u003cem\u003ePsychiatry, psychology, and law\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 194-205 (2019). https://doi.org/10.1080/13218719.2018.1487344\u003c/li\u003e\n\u003cli\u003eMoffitt, T. E. \u0026amp; Caspi, A. Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. \u003cem\u003eDevelopment and Psychopathology\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 355-375 (2001). https://doi.org/https://doi.org/10.1017/S0954579401002097\u003c/li\u003e\n\u003cli\u003eAzad, A. \u0026amp; Carlsson, J. Identity status and narrative identity processes in female adolescents\u0026apos; stories about committing crimes and being convicted. \u003cem\u003eJournal of adolescence (London, England.)\u003c/em\u003e \u003cstrong\u003e96\u003c/strong\u003e, 124-135 (2024). https://doi.org/10.1002/jad.12261\u003c/li\u003e\n\u003cli\u003ePulkkinen, L., Fadjukoff, P. \u0026amp; Pitk\u0026auml;nen, T. Persistent offenders and adolescence‐limited offenders: Differences in life‐courses. \u003cem\u003eCriminal Behaviour and Mental Health : CBMH\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 196-209 (2020). https://doi.org/https://doi.org/10.1002/cbm.2157\u003c/li\u003e\n\u003cli\u003eAzad, A. \u0026amp; Ginner Hau, H. Adolescent Females with Limited Delinquency: A Follow-Up on Educational Attainment and Recidivism. \u003cem\u003eChild \u0026amp; youth care forum\u003c/em\u003e \u003cstrong\u003e49\u003c/strong\u003e, 325-342 (2020). https://doi.org/10.1007/s10566-019-09530-8\u003c/li\u003e\n\u003cli\u003eBlokland, A. \u0026amp; van Os, R. Life Span Offending Trajectories of Convicted Dutch Women. \u003cem\u003eInternational criminal justice review\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 169-187 (2010). https://doi.org/10.1177/1057567710368938\u003c/li\u003e\n\u003cli\u003eMatsuda, M., Thornberry, T. P., Loughran, T. A. \u0026amp; Krohn, M. D. Are Late Bloomers Real? Identification and Comparison of Late-Onset Offending Patterns from Ages 14\u0026ndash;40. \u003cem\u003eJournal of Developmental and Life-Course Criminology\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 124-150 (2022). https://doi.org/https://doi.org/10.1007/s40865-022-00189-9\u003c/li\u003e\n\u003cli\u003eBroidy, L., Payne, J. \u0026amp; Piquero, A. R. Making Sense of Heterogeneity in the Influence of Childhood Abuse, Mental Health, and Drug Use on Women\u0026rsquo;s Offending Pathways. \u003cem\u003eCriminal Justice and Behavior\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 1565-1587 (2018). https://doi.org/https://doi.org/10.1177/0093854818776687\u003c/li\u003e\n\u003cli\u003eSkardhamar, T. DISTINGUISHING FACTS AND ARTIFACTS IN GROUP-BASED MODELING. \u003cem\u003eCriminology (Beverly Hills)\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, 295-320 (2010). https://doi.org/https://doi.org/10.1111/j.1745-9125.2010.00185.x\u003c/li\u003e\n\u003cli\u003eD\u0026apos;Unger, A. V., Land, K. C. \u0026amp; McCall, P. L. Sex Differences in Age Patterns of Delinquent/Criminal Careers: Results from Poisson Latent Class Analyses of the Philadelphia Cohort Study. \u003cem\u003eJournal of quantitative criminology\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 349-375 (2002). https://doi.org/10.1023/A:1021117626767\u003c/li\u003e\n\u003cli\u003eTheobald, D. \u0026amp; Farrington, D. P. in \u003cem\u003eEncyclopedia of Criminology and Criminal Justice\u003c/em\u003e (ed G. Bruinsma, Weisburd, D.) (Springer, 2014).\u003c/li\u003e\n\u003cli\u003eLangevin, S.\u003cem\u003e et al.\u003c/em\u003e Life-Course Persistent Antisocial Behavior and Accelerated Biological Aging in a Longitudinal Birth Cohort. \u003cem\u003eInternational journal of environmental research and public health\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 14402 (2022). https://doi.org/10.3390/ijerph192114402\u003c/li\u003e\n\u003cli\u003eZane, S. N., Welsh, B. C. \u0026amp; Zimmerman, G. M. Criminal Offending and Mortality over the Full Life-Course: A 70-Year Follow-up of the Cambridge\u0026ndash;Somerville Youth Study. \u003cem\u003eJournal of Quantitative Criminology\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 691-713 (2019). https://doi.org/https://doi.org/10.1007/s10940-018-9399-4\u003c/li\u003e\n\u003cli\u003eWickersham, A. \u003cem\u003eChanges in school performance and involvement in the criminal justice system\u003c/em\u003e, \u0026lt;https://www.adruk.org/fileadmin/uploads/adruk/Documents/Data_Insights/Data-Insight-Changes-in-School-Performance-Involvement-in-Criminal-Justice-System.pdf\u0026gt; (2024).\u003c/li\u003e\n\u003cli\u003eHunter, K. The time is now: Preventing unnecessary criminalisation of children in care. \u003cem\u003eInternational journal of population data science\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2023). https://doi.org/10.23889/ijpds.v8i2.2193\u003c/li\u003e\n\u003cli\u003eMinistry of Justice. Female Offending Blueprint for Wales. (2019).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8115965/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8115965/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"As females are a minority group in the criminal justice system, they have often been neglected in criminological research. Research exploring offending trajectories has largely focused on male and mixed-gender samples and studies using female-only cohorts have been constrained by small sample sizes, short follow-ups and limited evidence from England. However, as rates of female imprisonment increase, it is ever more important that we attempt to understand females’ engagement with the criminal justice system. This study used a linkage of the Police National Computer and the National Pupil Database to create a birth cohort of female offenders in England (N=196,089). The study explored their offending trajectories from the age of criminal responsibility (10 years) up to age 31 years. A latent class analysis identified a four-class model with the following classes: ‘Life-course-persistent’ (11%), ‘Adolescent-limited’ (54%), ‘Young-adult-limited’ (16%) and ‘Adult-onset’ (19%). These classes resemble those identified using male and mixed-gender samples. By identifying the key developmental patterns of female offending in England, this study provides an empirical basis for early identification and tailored support for those females most at risk of persistent offending.","manuscriptTitle":"Female reoffending trajectories in England from 2000 to 2020: A longitudinal administrative data study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-25 07:12:46","doi":"10.21203/rs.3.rs-8115965/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"307d9089-17ec-49af-9e3a-1743d599c7d0","owner":[],"postedDate":"November 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":58333635,"name":"Social science/Criminology"},{"id":58333636,"name":"Biological sciences/Psychology"},{"id":58333637,"name":"Social science/Psychology"}],"tags":[],"updatedAt":"2025-11-25T07:12:46+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-25 07:12:46","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8115965","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8115965","identity":"rs-8115965","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}