Childhood predictors of successful self-reported delinquents

ABSTRACT

The main aim of this research is to investigate the childhood predictors of successful self-reported delinquents, defined as those who were not convicted. In the Cambridge Study in Delinquent Development (CSDD), 411 London males have been followed up from age 8 to age 61. Self-reported offending was measured for the whole sample for ages 10–14, 15–18, 27–32, and 42–47, for five crimes: burglary, theft of a vehicle, theft from a vehicle, shoplifting, and vandalism. The prevalence of self-reported offending was 64% at ages 10–18 and 11% at ages 27–47, compared with the prevalence of convictions of 23% and 8% respectively. Successful self-reported delinquents were defined as those who offended between ages 10 and 18 but were not convicted up to age 26; 126 successful delinquents were compared with 120 convicted delinquents. Sixteen childhood factors, including attainment, self-control, socioeconomic, parental, family and behavioural factors, predicted successful self-reported delinquents. The most important independent predictors were committing less serious and fewer offences as well as high school attainment, unconvicted parents, low risk-taking, and unseparated families.

KEYWORDS:

• Successful delinquents
• self-reported offences
• convictions
• longitudinal study
• childhood factors

Introduction

Successful offenders

‘Success’ in criminology has typically been considered from the viewpoint of re-establishing life success after a person has committed an offence. This process has been detailed by a number of researchers (e.g. Maruna, 2001; Sampson & Laub, 1993) who have documented how former offenders can become more successful in their lives (e.g. by gaining employment, maintaining stable relationships) while transitioning out of offending. Similarly, using the Cambridge Study in Delinquent Development (CSDD), which is a prospective study of 411 London boys followed up in interviews from ages 8 to 48, Farrington et al. (2006) showed that convicted males who were no longer convicted after age 21 (desisters) were as successful in their lives as males who had no convictions. Based on the same study, and using conviction data up to age 61, Farrington (2019a) showed that the probability of being reconvicted decreased with increasing time since the previous conviction, but that this was still substantial (19%) even after 15 crime-free years.

A much less studied, but equally important operationalisation of ‘success’ in criminology, is being successful at crime. This might include maximising the income generated from offending, making ‘easy money’ illegally or being skilled at avoiding detection and beating the system (e.g. Brezina & Topalli, 2012). This form of success is sensationalised in the autobiographies of high-profile criminals (e.g. Pearson, 2015) but it can also be found in interviews with specific groups such as drug dealers (e.g. Jacobs, 1996) or in ethnographic research into grey areas of criminal behaviour such as ticket touting (Moretti, 2023). These contributions provide interesting and important contextual information about decision making on offending but suffer the limitations of retrospective recall bias and the risk of demand characteristics. The agreement between prospective and retrospective self-reports of offending is often poor (e.g. Baldwin et al., 2019; Kazemian & Farrington, 2005).

Another source of information about criminal success comes from surveys of those in prison. Brezina and Topalli (2012) analysed the results of interviews that had been conducted with 700 male prisoners in Nebraska prisons in 1989–1990. The interviews covered topics such as their level of past criminal involvement, the income that they had generated from crime and their past experiences with arrests and incarceration. The results showed that, controlling for age, ethnicity, education level, marital status, and the number of self-reported offences and arrests, a low ratio of arrests to self-reported offences predicted how successful these individuals felt they were at crime.

Chaiken and Chaiken (1985) used a survey of over 11,000 male prison inmates from California, Michigan and Texas to identify the factors that were associated with a group they termed ‘high-rate winners’. High-rate winners were those who were not arrested frequently, but who reported committing offences at rates higher than 70% of the inmates who also reported committing these offences. Two reasons were proposed for the success of high-rate winners. The first reason was that, even though they were similar in their illicit drug use to others, they were less likely to use heroin. It was proposed that this helped these offenders to avoid tell-tale appearances such as track marks, ulcers or wearing inappropriate clothing during hot weather to cover up physical signs of addiction. The second reason was that high-rate winners were more careful planners, they were more likely to follow victims to safe places before robbing them, and they were more likely to commit offences with partners.

In a prospective study of 124 convicted French-Canadian males followed up from about age 15 to their early 40 s, Kazemian and Le Blanc (2007) examined the individual and social characteristics that predicted ‘differential cost avoidance’. This was a composite measure that combined the ratio of self-reported career length to officially recorded career length, the ratio of self-reported offending gravity to officially recorded gravity, and the ratio of time free to periods of incarceration. The results suggested that the main predictors of criminal success in the 30s were not using substances (especially drugs), having debts, and not using or threating violence in the perpetration of offences.

Why are some self-reported offenders never convicted?

There are a number of imperfect methods of developing a picture of the, so called ‘dark figure’ of crime, or the totality of crime that is committed which does not necessarily come to the attention of police (e.g. Biderman & Reiss Jr., 1967; Skogan, 1977). These include surveys asking about victimisation (e.g. Office of National Statistics, 2022), and monitoring A&E admissions (e.g. Sivarajasingam et al., 2003) amongst others, but the most commonly used approach is asking individuals to self-report the offences that they have committed (Krohn et al., 2010).

A vast amount of research has shown that people self-report many more offences than are found in official conviction records (e.g. Farrington, 1989; Farrington et al., 2013a; Gomes et al., 2018; Theobald et al., 2014; West & Farrington, 1977). However, important variation in the correspondence of these two measures of offending has been noted (e.g. Krohn et al., 2010; Thornberry & Krohn, 2000). This may be attributable to methodological issues, such as the wording of certain self-reported items not corresponding well to the official definition of an offence and/or the method of administration of the self-reported measure (i.e. anonymous versus non-anonymous; Gomes et al., 2018). However, this may also reflect systemic bias. For example, it is argued that the criminal justice system was established and oriented towards funnelling individuals with certain characteristics (i.e. young, Black and poor) into the system and progressing them further (Lammy, 2017). While some research has shown that the correspondence between self-reports and official records vary between those of different ethnicities (i.e. Jolliffe et al., 2003), it is generally accepted that self-reports are a valid measure of offending, or at least as valid as any other measure available (Gomes et al., 2018; Jolliffe & Farrington, 2014; Krohn et al., 1974, 2010, 2013; Thornberry & Krohn, 2000).

Another, but much more rarely investigated possibility for the variation in the correspondence between self-reports and official records is that an offence is committed but it is not officially identified. One potential reason for this is that those who are not caught are good at avoiding detection by authorities, for example through their choice of victims. Some victims (e.g. sex workers) are less likely to report their victimisation to the police, compared to others (Lee & Reid, 2018), adding to the ‘dark figure of crime’ (e.g. Biderman & Reiss, 1967; Skogan, 1977).

An offence may also not translate into a conviction if police or prosecutors decide that it is not in the public interest to pursue a conviction because the offence was not considered sufficiently serious (e.g. being drunk in a public place; CPS, 2022). Early research on the ‘dark figure of crime’ supports this notion (Skogan, 1977). Even if the decision to prosecute is taken, the trial could be ‘cracked’ (i.e. on the trial date the defendant offers acceptable pleas or the prosecution offers no evidence) or the defendant could be found not guilty. In the UK, the rate of unsuccessful conviction was reported to be about 17% in 2022 (Ministry of Justice, 2022).

Police discretion may also influence correspondence. That is, a high level of antisocial and criminal behaviour is common in adolescence (e.g. Moffitt, 1993), and as a result the authorities might choose to show discretion with some young people. It is also possible that there might be individual and family level factors that contribute to the authorities taking a more lenient stance with some young people. For example, it could be that the young person uses manipulation and deceitfulness to convince the authorities not to charge them, and thus navigates out of the system. It could also be that certain family features could protect a young person from a charge or conviction. This may be features such as having parents who are able and willing to support the young person and convince the authorities to use discretion. This would avoid the reduced life opportunities which a conviction brings (e.g. Petrosino et al., 2019). Furthermore, diversion may reduce costs as young people who commit less serious offences tend to desist naturally as a result of maturity and age (Moffitt, 1993).

Very little empirical research has investigated why young people might engage in offending (as measured by self-report; SR), but never receive a conviction; or ‘successful self-reported delinquents’. One study that has attempted to investigate this is by Youngs and Canter (2014). They investigated group differences in offending between unmatched convicted (n = 185) and non-convicted (n = 80) SR delinquents. Both groups were asked to self-report whether they had committed a wide range of offences (45 types ranging from minor offences such as shoplifting to more serious offences such as stealing a car, burglary, and murder). The results indicated that the prevalence of self-reported offending was higher amongst the convicted versus the non-convicted group.

Youngs and Canter (2014) also attempted to identify the aspects that differentiated successful offenders (non-convicted SR offenders) from non-successful offenders (convicted SR offenders). By comparing the prevalence and rankings of the offences in the two groups they concluded that ‘instrumentality’ was what separated the two groups. That is, those who had been convicted tended to use criminal behaviour as a tool to reach a goal or gain a material need, compared to those who were not convicted. However, Youngs and Canter (2014) did not investigate psychosocial predictors of successful self-reported offenders.

Theobald et al. (2014) investigated the relationship between self-reported offending and official convictions in the oldest cohort (n = 506) of the Pittsburgh Youth Study (PYS) between ages 13 and 24. The offences studied were theft (serious/moderate) and violence (serious/moderate). The number of self-reported offences in this sample across adolescence was over 22,341 (49.21 per boy), while the cohort only had 1,136 convictions (2.26 per boy). Those who self-reported theft and violence that was less serious were more likely to be undetected (45.6% and 34.1%, respectively), compared to those who self-reported more serious theft (25.2%) or violence (21%). Unsurprisingly, given the well-established link between committing serious offences (particularly violence) and the frequency of offending (i.e. Farrington, 1998), those who self-reported a greater number of offences were significantly more likely to be detected. Only 38% of those who self-reported between one and four offences had a conviction, compared to over 72% of those who self-reported 23 or more offences. The results showed that a significant number of SR offenders went undetected. Out of the 323 boys who self-reported at least one offence, 150 (or 46.4%) did not have a conviction. However, White SR delinquents were significantly less likely to be convicted than African American SR delinquents (42% versus 62%).

The Seattle Social Development Project (SSDP), a longitudinal study of 808 participants followed from childhood into early adulthood was also used to investigate the relationship between self-reports and official offending (Gilman et al., 2014). As expected, self-reported prevalence and frequency of offending was much higher than that of official charges. At age 21, the prevalence of any SR offence was 52.8%, compared to 6.2% for official offences, while the mean frequency of SR offending was about 23 offences per offender compared to 2 for official offences. Those who self-reported violent offences tended to have the highest probability of being charged (ranging from 26.9% to 57.5%), followed by self-reported property offenders (ranging from 18.4% to 25.0%). Drug offenders were the least likely to be charged (3.4% to 6.0%). Examining the independent predictors of an official charge Gilman et al. (2014) found that the frequency of self-reported offending was significantly related for all of property, violent and drug offending as was poverty (property, violence), being male (violent and drug) and being African American (violent and drug).

It is perhaps not surprising that research has shown that those who self-report violence are comparatively more likely to receive a corresponding charge. Violence is a serious interpersonal offence so the police may be more likely to allocate resources into the detection and prosecution of these offenders. Also, someone who commits violence is more likely to have been seen compared to someone who has committed burglary, for example, and therefore there is an increased likelihood of identification. Last, as previously mentioned, longitudinal research has shown that those who commit violence are much more likely to be frequent offenders (i.e. Farrington, 1998), and those who commit many offences are more likely to eventually end up with a charge.

Identifying likely predictors of successful self-reported delinquents

Clearly, there are a number of potential factors that might increase or reduce the likelihood of becoming a successful SR delinquent. To the authors’ knowledge, there are no studies to date that have systematically investigated the psychosocial predictors of successful SR offenders. There have however, been a number of studies that have identified protective factors, defined as those factors that predict a low likelihood of involvement in offending generally. Lösel and Farrington (2012) reviewed the literature on protective factors and found that being intelligent, not being daring or impulsive, having positive family relationships, attending a good school and having a good relationship with the school, high school attainment, having prosocial friends, high socio-economic status (SES), and living in a good area/neighbourhood were all associated with lower levels of offending. Additional research (e.g. Dubow et al., 2016; Farrington et al., 2016; Jolliffe et al., 2016; Kim et al., 2016) has identified similar factors, but the extent to which these predict ‘successful’ self-reported offenders is unknown. Addressing this question requires prospective longitudinal studies which assess these factors before any offending takes place (to establish temporal ordering) and which collect concurrent self-reports and official records over a long enough period of time for a conviction to be recorded.

Current study

The current study aims to identify the factors that predict those who self-report offences but are never convicted, or the characteristics of successful SR offenders. Based on the limited literature, there is evidence that those who are less frequent and less serious in their offending may be more likely to be successful (Gilman et al., 2014; Theobald et al., 2014; Youngs & Canter, 2014), but there may be many other factors. As a result, the current study’s main aim is to investigate childhood predictors of successful SR delinquents. The following research questions are important:

1. What is the relationship between self-reported delinquency and convictions, and how many people are identified as successful SR delinquents?

2. How are the seriousness and frequency of self-reported offending related to the probability of convictions?

3. Which childhood psychosocial factors are associated with becoming a successful SR delinquent?

4. Controlling for offence seriousness and frequency, which childhood psychosocial factors independently predict successful SR delinquents?

Method

Sample

The CSDD is a prospective longitudinal survey of 411 London males who were first studied in 1961–1962 at age 8–9. Their parents, teachers, peers, female partners, and children have also been interviewed. At the time that they were first contacted in 1961–1962, the males were all living in a working-class area of South London. The vast majority of the sample was chosen by taking all the males who were then aged 8–9 and on the registers of six state primary schools within a one-mile radius of a research office which had been established. Most boys were born in 1953. In addition to 399 males from these six schools, 12 males from a local school for ‘educationally subnormal’ (special needs) children were included in the sample, in an attempt to make it more representative of the population of males living in the area. Therefore, the males were not a probability sample drawn from a population, but rather a complete population of males of that age in that area at that time.

Most of the males (357, or 87%) were White in appearance and of British origin, in the sense that they were being brought up by parents who had themselves been brought up in England, Scotland, or Wales. Of the remaining 54 males, 12 were African-Caribbean, having at least one parent of West Indian (usually) or African origin. Of the remaining 42 males of non-British origin, 14 had at least one parent from the North or South of Ireland, 12 had parents from Cyprus, and the other 16 males were White and had at least one parent from another Western industrialised country.

On the basis of their fathers’ occupations when they were aged 8, 94% of the males could be described as working-class (categories III, IV, or V on the Registrar General’s scale, describing skilled, semi-skilled or unskilled manual workers), in comparison with the national figure of 78% at that time. The majority of the males were living in conventional two-parent families with both a father and a mother figure; at age 8, only 6% of the males had no operative father and only 1% had no operative mother. This was, therefore, overwhelmingly a traditional White, urban, working class sample of British origin.

The males have been interviewed nine times, at ages 8, 10, 14, 16, 18, 21, 25, 32, and 48. At all ages except 21 and 25, the aim was to interview all the males who were still alive,¹ and it was always possible to interview a high proportion: 405 (99%) at age 14, 399 (97%) at age 16, 389 (95%) at age 18, 378 (94%) at age 32, and 365 (93%) at age 48. The interviews were approved by various ethics committees, including those of the Cambridge Institute of Criminology, the Home Office, and the Institute of Psychiatry, Kings College London.

The results of the CSDD have been described in six books (Farrington et al., 2013b; Piquero et al., 2007; West, 1969, 1982; West & Farrington, 1973, 1977), and in nine summary articles (Farrington, 1995, 2003, 2019a, 2021; Farrington et al., 2009b, 2021; Farrington & Jolliffe, 2022; Farrington & West, 1981, 1990).

Self-reported offending

Eight types of crimes were enquired about at the main ages of 14, 18, 32, and 48, when most males were interviewed: burglary, theft of a vehicle, theft from a vehicle, shoplifting, vandalism, theft from a slot machine, assault, and drug use. The reference periods were ‘ever’ (age 14), last three years (age 18), and last five years (ages 32 and 48). These time periods did not overlap, but the questions at age 16 (ever) did overlap with the age 14 and 18 questions and so were not studied here. Also, theft from a slot machine was not studied here because there were hardly any convictions for this crime at older ages (see Farrington et al., 2013a). Drug use was not included because this is commonly considered a ‘victimless’ crime, it has been decriminalised in many jurisdictions, and previous research has shown this to be a high frequency offence with very low detection rates (e.g. Farrington et al., 2003; Gilman et al., 2014). This would have the impact of having regular drug users influencing who were considered successful offenders.

Despite violence being an important factor associated with the likelihood of conviction (e.g. Gilman et al., 2014), assault was not included here. This was because the self-report items for assault mainly referred to being involved in a fight, and it was not clear that this behaviour would have led to a conviction for assault. Also, it was very prevalent (e.g. 64.5% at age 18). Second, all other offences where self-reported data were available were property offences, so including assault would introduce unneeded heterogeneity. This left five crimes with reasonably comparable self-report and conviction measures at different ages: burglary and theft of a vehicle (relatively serious crimes) and theft from a vehicle, shoplifting, and vandalism (relatively minor crimes). According to the most recent Home Office estimates (Heeks et al., 2018), domestic burglary (£5,930), commercial burglary (£15,460) and theft of a vehicle (£10,290) are much more costly, and therefore more serious, than theft from a vehicle (£870), vandalism (£1,350) and shoplifting (£970 for commercial theft). However, there may be within-crime variations in seriousness, and it is possible that different results might be obtained with a different metric of seriousness.

The exact wording of the self-report items at different ages is shown in Farrington (1989). Briefly, burglary usually specified breaking and entering and then stealing something. Theft of a vehicle usually specified driving a car, van, motorcycle, scooter (etc.) that had been taken without the owner’s permission. Theft from a vehicle usually specified stealing from parked cars, vans, trucks (etc.). Shoplifting usually specified shoplifting from shops, market stalls, stores, supermarkets (etc.). Vandalism usually specified deliberately damaging property, such as telephone boxes, cars, windows (etc.) without stealing anything.

Criminal record searches

The criminal records of the males have been searched repeatedly since 1964. The most recent search was in April 2017 (see Farrington, 2019b). Only relatively serious offences are recorded on the Police National Computer (PNC), for example excluding almost all motoring offences, and only crimes committed on different days were counted. The minimum age of conviction is 10. The age at which the offence was committed was counted, not the age at the time of the conviction. Given the typical delay of several months between the commission of an offence and a conviction appearing in the PNC, it is likely that all offences committed up to the end of August 2016 were recorded. Since the youngest male in the CSDD was born in August 1954, it is considered that all males have been searched in criminal records at least up to their 62nd birthday. Therefore, information is available about all offences committed up to age 61.99.

For comparability with self-reports, convictions at ages 10–14, 15–18, 27–32, and 42–47 were studied. The mean age at the age 48 interview was 48.0, and more than half of the males were age 47, so the most relevant five-year period was considered to be 42–47.

Childhood psychosocial factors

At ages 8–10, the aim was to measure as many variables as possible that were believed (at the time) to predict offending. There was a concern that too many criminological studies measured and analysed a very limited number of variables. In many CSDD analyses, the age 8–10 variables have been dichotomized into the ‘worst’ quarter versus the remainder. This facilitated a ‘risk factor’ approach, made all the risk factors comparable, and did not usually involve much loss of information, as many variables were originally measured on 2, 3, or 4 point scales (Farrington, 2020; Farrington & Loeber, 2000). They were not measured on normally distributed equal-interval scales.

As the present article is concerned with the prediction of successful SR delinquents, the positive ends of variables were studied as well as the negative ends. Loeber et al. (2008) proposed that a variable that predicted a low rate of offending should be termed a ‘promotive’ factor, in order to distinguish it from ‘protective’ factors, which were often defined as variables that interacted with risk factors in order to nullify their effects (e.g. Farrington & Bergstrøm, 2021). In order to investigate risk and promotive factors in the Pittsburgh Youth Study, Loeber et al. (2008) trichotomized variables into the ‘worst’ quarter (e.g. low school achievement), the middle half, and the ‘best’ quarter (e.g. high school achievement). ‘Promotive’ factors have often been termed ‘direct protective’ factors, to distinguish them from ‘interactive protective’ factors, and we will use this terminology in this article (Farrington et al., 2016; Lösel & Farrington, 2012).

Trichotomized variables were used in the CSDD by Farrington and Ttofi (2011), who investigated the extent to which variables measured at ages 8–10 predicted convictions up to age 50. The present article uses trichotomized variables studied in a later article by Farrington et al. (2016), who investigated the extent to which these variables predicted convictions between ages 10 and 18. Farrington et al. (2016) listed 24 trichotomized variables. Of these, 4 were excluded because of small numbers in a category: nervousness of the boy, number of friends of the boy, socioeconomic status, and interest in education by the parents. Of the remaining 20 trichotomized variables that were studied, 6 were not significantly related to successful self-reported offenders: extraversion, neuroticism, popularity, nervousness of the mother, job of the mother and child rearing. Excluding these left the 14 trichotomized variables shown in Table 3, plus two dichotomous variables (housing and convicted parent). The 16 childhood (ages 8–10) variables studied here are as follows; they were all significant predictors of successful SR delinquents.

Attainment

The non-verbal IQ of the boy was measured by the Progressive Matrices test, while verbal IQ was assessed based on verbal comprehension and vocabulary tests. Junior school attainment of the boy was based on school records of arithmetic, English and verbal reasoning tests.

Self-control

Daring was rated by parents and peers, and identified boys who took many risks in traffic, climbing, exploring, etc. Hyperactivity was based on ratings by teachers of whether the boy lacked concentration or was restless in class, while impulsiveness was based on psychomotor tests of clumsiness.

Socio-economic

Family income, housing, and family size (number of children) were rated by the psychiatric social workers who interviewed the boys’ parents. Housing was only coded as a dichotomous variable (tolerable or poor). Information about the delinquency rates of schools was obtained from the local education authority.

Parental

Information about whether the father and/or mother were convicted up to the boy’s tenth birthday was obtained from criminal record searches. This was a dichotomous variable. The age of the mother was rated by the psychiatric social workers. Young mothers (the risk category) were those who had their first child before age 20, while older mothers (the promotive category) were those who had their first child at age 25 or older.

Family

Parental supervision was rated by the psychiatric social workers, and this referred to the extent to which the parents knew what the boy was doing when he was outside the house. A disrupted family referred to temporary or permanent separations of a boy from a parent before the boy’s tenth birthday for reasons other than death or hospitalisation.

Behaviour

Dishonesty was rated by peers, while troublesomeness was rated by peers and teachers, identifying boys who got into trouble most.

Results

Prevalence and frequency of offending

Table 1 shows the prevalence and frequency of offending at different ages. Altogether, 387 males were interviewed at both 14 and 18, while 408 males were known on convictions between ages 10 and 18 (excluding three males who emigrated up to age 14). A total of 246 males (63.6%) admitted at least one offence at these ages, including 109 (28.2%) who admitted a serious crime, and 230 (59.4%) who admitted a minor crime. Regarding convictions, 93 males (22.8%) were convicted of at least one of these offences between ages 10 and 18, including 70 (17.2%) who were convicted of a serious crime, and 47 (11.5%) who were convicted of a minor crime. There were 2.8 self-reported offenders for every convicted offender, but only 1.6 serious self-reported offenders for every serious convicted offender, and 5.2 minor self-reported offenders for every minor convicted offender.

Table 1. Prevalence and frequency of offending.

	Ages 10–14, 15–18			Ages 27–32, 42–47
	SR (387)	Con (408)	Ratio	SR (390)	Con (393)	Ratio
Prevalence
Burglary	19.9	11.0	1.8	2.3	2.8	0.8
Theft of vehicle	20.2	12.0	1.7	2.8	1.8	1.6
Theft from vehicle	19.6	5.1	3.8	2.6	1.0	2.6
Shoplifting	46.8	4.9	9.6	6.2	2.5	2.5
Vandalism	28.9	2.5	11.6	1.8	2.5	0.7
Serious SR crime	28.2	17.2	1.6	4.1	3.8	1.1
Minor SR crime	59.4	11.5	5.2	8.5	5.1	1.7
All 5 crimes	63.6	22.8	2.8	11.3	7.6	1.5
Frequency
Burglary	677	77	9.3	18	11	1.6
Theft of vehicle	533	64	8.8	118	8	14.9
Theft from vehicle	601	31	20.4	132	4	33.3
Shoplifting	2130	26	86.4	422	11	38.7
Vandalism	795	11	76.2	7	11	0.6
Serious SR crime	1210	141	9.0	136	19	7.2
Minor SR crime	326	68	54.7	561	26	21.7
All 5 crimes	4736	209	23.9	697	45	15.6

Notes: SR = self-report; con = conviction; Ratios of frequencies takes account of Ns.

Turning to frequency between ages 10 and 18, these males admitted 4,736 crimes in total, including 1,210 serious crimes and 3,526 minor crimes. In contrast, there were 209 convictions in total, including 141 serious convictions and 68 minor convictions. The self-reported frequency rate per male was 12.24 offences (4,736/387), compared with the conviction frequency rate of 0.51 (209/408) per male, yielding a ratio of 23.9 self-reported offences per conviction. The ratio was 9.0 for serious crimes and 54.7 for minor crimes.

At the adult ages, the 390 males who were interviewed at either or both of ages 32 and 48 were studied for self-reports. In total, only 44 males (11.3%) admitted at least one of these offences, including 16 (4.1%) who admitted a serious crime and 33 (8.5%) who admitted a minor crime. Regarding convictions, 393 males were known (i.e. they were not dead or emigrated) in at least the first of the two age ranges (27–32 and 42–47). Of these males, 30 (7.6%) were convicted during at least one of these age ranges, including 15 (3.8%) convicted of a serious crime and 20 (5.1%) convicted of a minor crime. The self-report to conviction ratio was only 1.5 for all crimes, 1.1 for serious crimes, and 1.7 for minor crimes.

At the adult ages, these males admitted 697 crimes in total, including 136 serious crimes and 561 minor crimes. In contrast, there were 45 convictions in total, including 19 serious convictions and 26 minor convictions. The self-reported frequency rate per male was 1.79 offences (697/390), compared with the conviction frequency rate of 0.11 (45/393), yielding a ratio of 15.6 self-reported offences per conviction. The ratio was 7.2 for serious crimes and 21.7 for minor crimes.

Because of the very small number of self-reported and convicted offenders at the adult ages, the remainder of this study of successful SR delinquents focuses on ages 10–18.

Juvenile self-reported offending versus convictions

Table 2 shows the relation between self-reported offending and convictions at ages 10–18. For the five specified crimes, 246 males admitted at least one offence and 79 of these males (32.1%) were convicted of at least one offence. The remaining 167 males could be viewed as successful SR delinquents. Of the 141 males who denied all five crimes, 8 (5.7%) were convicted of at least one. Each of these eight males had only one relevant conviction (for any of the specified crimes). Not all were under-reporting. In particular, one male was convicted of vehicle theft, but the court report showed that he was only carried in a stolen vehicle. Therefore, he would not have said yes to our item, which specified driving a stolen vehicle. Two other males were convicted of damaging property; one broke fencing at a pop festival and used it for firewood, while the other cut string to release bales of hay. Neither would necessarily admit an item that specified ‘deliberately damaging property, such as telephone boxes, cars, windows, etc. (without stealing anything)’.

Table 2. Juvenile self-reported offending versus convictions.

	5 crimes			Con 10–18			Con 10–26
	NC	C	% C	NC	C	% C	NC	C	% C
SR crime
No	133	8	5.7	127	14	9.9	116	25	17.7
Yes	167	79	32.1	147	99	40.2	126	120	48.8
	(OR = 7.86)			(OR = 6.11)			(OR = 4.42)
SR serious crime
No	256	22	7.9	238	40	14.4	212	66	23.7
Yes	44	65	59.6	36	73	67.0	30	79	72.5
	(OR = 17.19)			(OR = 12.07)			(OR = 8.46)
SR minor crime
No	140	17	10.8	133	24	15.3	122	35	22.3
Yes	160	70	30.4	141	89	38.7	120	110	47.8
	(OR = 3.60)			(OR = 3.50)			(OR = 3.20)
No. SR crimes
0	133	8	5.7	127	14	9.9	116	25	17.7
1–3	65	10	13.3	64	11	14.7	56	19	25.3
4–9	51	15	22.7	43	23	34.8	40	26	39.4
10–29	38	18	32.1	31	25	44.6	24	32	57.1
30+	13	36	73.5	9	40	81.6	6	43	87.8
	(x^2= 102.56)			(x^2= 105.63)			(x^2= 90.40)

Note: C, con = conviction, NC = Not convicted, SR = Self-reported, all significance tests p < .0001.

Of the other five males, three were convicted of shoplifting, one of burglary, and one of theft from a vehicle. Based on the court records, they should have admitted these crimes. Therefore, it seems that five of the non-admitters (3.5%) were under-reporting (either forgetting or deliberately concealing).

In order to be more cautious about identifying successful SR delinquents, the males who were convicted of any offence between ages 10 and 18 were studied. In total, 120 males were convicted between ages 10 and 18, of whom 93 (77.5%) were convicted of one of the five specified offences. Table 2 shows that, of the 246 SR delinquents, 99 were convicted of any offence, leaving 147 successful SR delinquents.

In order to be even more cautious, the time period for convictions was extended up to age 26. By this age, 152 males were convicted of any offence, 85.4% of the total 178 who were convicted up to age 61. Table 2 shows that, of the 246 SR delinquents, 120 were convicted of any offence up to age 26, compared with 126 who were not convicted. In the remaining analyses, these 126 successful SR delinquents will be compared with the 120 convicted SR delinquents.

Table 2 also shows that serious SR delinquents were much more likely to be convicted up to age 26 than minor SR delinquents (72.5% compared with 47.8%). Similarly, the probability of conviction increased with the number of these crimes committed, from 17.7% of those who committed none of these five crimes at ages 10–18 to 87.8% of those who committed 30 or more crimes.

Predicting successful SR delinquents

Table 3 shows the extent to which the childhood factors at ages 8–10 predicted the 126 successful SR delinquents compared with the 120 convicted delinquents. For example, of 53 males with a high non-verbal IQ, 32 (60.4%) were successful, compared with 71 of 127 males with a medium non-verbal IQ (55.9%), and 23 of 66 males with a low non-verbal IQ (34.8%); chi-squared = 9.97 with 2 degrees of freedom, p = .003 (one-tailed tests used because of clear directional predictions).

Table 3. Predicting successful self-reported delinquents.

	% Successful (51.2%)			p values
	Prot	Med	Risk	Succ	Minor	No. SR
Attainment
Nonverbal IQ	60.4	55.9	34.8	.003	.0003	.044
Verbal IQ	73.1	49.2	38.6	.0004	.003	.031
School attainment	67.9	53.6	27.9	.0001	.0005	NS
Self-control
Daring	73.1	50.5	39.1	.0003	.0003	.003
Hyperactivity	59.4	48.8	39.7	.023	.002	.046
Impulsiveness	61.1	54.8	36.8	.008	NS	NS
Socioeconomic
Family income	61.9	54.2	35.4	.004	NS	NS
Housing	57.2	–	41.5	.008	.007	NS
Family size	59.4	58.9	31.4	.0003	.007	NS
School delinquency	67.6	45.2	32.7	.0002	.0003	NS
Parental
Convicted parent	60.8	–	29.3	.0001	.0001	.001
Age of mother	60.7	52.4	39.3	.029	.007	NS
Family
Supervision	65.3	53.5	35.3	.005	.021	NS
Separations	60.0	45.7	35.0	.002	NS	NS
Behaviour
Dishonesty	68.8	54.1	35.8	.001	.0001	.029
Troublesomeness	75.7	50.9	26.5	.0001	.0001	.0001
Minor/serious SR	70.1	–	27.5	.0001	–	.0001
1-5/6+ SR	76.1	–	30.1	.0001	.0001	–

Notes: SR = Self-Report; Prot = Direct protective category; Mid = Middle category; Risk = Risk category, Succ = p values for predictors of successful versus convicted SR delinquents; Minor = p values for predictors of minor versus serious SR delinquents. No. SR = p values for predictors of 1–5 versus 6+ SR crimes, NS = Not Significant.

The 16 significant age 8–10 variables are shown in Table 3. It is remarkable how many childhood factors in six categories (attainment, self-control, socio-economic, parental, family, and behaviour) predicted successful SR delinquents. The most significant trichotomized variable was troublesomeness. Of 70 males with low troublesomeness, 53 (75.7%) were successful, compared with 55 of 108 males with medium troublesomeness (50.9%), and 18 of 68 males with high troublesomeness (26.5%); chi-squared = 33.48; p < .0001.

Table 3 also shows relationships with the seriousness and frequency of SR offending. For example, of 137 only minor offenders, 96 were successful (70.1%), compared with 30 of 109 serious offenders (27.5%); chi-squared  = 43.99, 1 df, p < .0001. The number of SR offences was dichotomized at 1–5 versus 6 or more, because those committing 6 or more offences seemed qualitatively different from those committing 1–5 offences. Table 2 shows that 25.3% of those committing 1–3 crimes were convicted, compared with 39.4% of those committing 4–9 crimes, and 57.1% of those committing 10–29 crimes. Further investigation showed that 8 out of 38 of those committing 4–5 crimes (21.1%) were convicted, compared with 18 out of 28 of those committing 6–9 crimes (64.3%). Therefore, the 4–5 males were similar to the 1–3 males, and the 6–9 males were similar to the 10–29 males, justifying the dichotomization at 1–5 versus 6 or more. Of 113 males with 1–5 offences, 86 were successful (76.1%), compared with 40 of 133 males (30.1%) with 6 or more offences; chi-squared = 51.81, 1 df, p < .0001.

Perhaps not surprisingly, minor offending was related to infrequent offending. Of 137 only minor offenders, 84 were infrequent offenders, with 1–5 offences (61.3%), compared with 29 of 109 serious offenders (26.6%); chi-squared = 29.45, 1 df, p < .0001.

Table 3 also shows relations between the 16 childhood variables and minor and infrequent offending. Low troublesomeness was the variable that was most strongly related to both of these measures. Three variables were not significantly related to minor offending, while nine variables were not significantly related to infrequent offending.

Independent predictors of successful SR delinquents

In order to investigate independent predictors of successful SR delinquents, a series of logistic regressions were carried out, using the forward stepwise method (in order to minimise the problem of multicollinearity). Minor and infrequent SR offending were included in each regression, because of their importance as predictors; they always entered the equation first in stepwise regressions. Variables in the six categories shown in Table 3 were entered in six separate regressions. For example, non-verbal IQ, verbal IQ, and school attainment were entered in the first regression, along with minor and infrequent offending. Infrequent offending entered the equation first (p < .0001), followed by minor offending (p < .0001), followed by school attainment (p = .001). However, non-verbal IQ and verbal IQ were not significant. In the other five analyses, the following childhood variables were the best independent predictors: daring (p = .012), school delinquency (p = .002), convicted parent (p = .003), separations (p = .001), and troublesomeness (p = .004).

These six variables were then entered into a final logistic regression analysis, along with minor and infrequent self-reported offending. Table 4 shows the results. School attainment, convicted parent, daring, and separations were significant independent predictors of successful SR delinquents, but troublesomeness and school delinquency were not. Therefore, the main conclusion is that boys with high school attainment, boys who do not take many risks, boys with unconvicted parents, and boys from unbroken two-parent families were the most likely to become successful SR delinquents.

Table 4. Results of logistic regression analysis.

	LRCS	p	B	SE	p
1-5/6+ SR	45.29	.0001	1.57	0.36	.0001
Minor/serious SR	22.57	.0001	1.22	0.36	.0004
School attainment	11.39	.0004	0.65	0.26	.005
Convicted parent	6.42	.006	0.78	0.41	.030
Daring	4.56	.016	0.47	0.24	.023
Separations	3.21	.037	0.39	0.22	.038

Notes: LRCS = Likelihood Ratio Chi-Squared change; SE = Standard Error; SR = Self-Report.

Did not enter the equation: Troublesomeness, school delinquency.

Discussion

Self-reported delinquency versus convictions

The aim of this study was to identify successful SR delinquents, or those who reported committing offences but were not convicted, and to determine whether these individuals could be predicted based on psychosocial factors measured in childhood. In this study 126 males self-reported committing at least one of five types of offences (burglary, theft of a vehicle, theft from a vehicle, shoplifting, vandalism) between the ages of 10 and 18 and had no convictions up to age 26, so were considered successful. Comparatively, 120 SR delinquents were convicted. A number of childhood factors in the domains of attainment, self-control, socio-economic, parental, family and behaviour predicted successful offending.

From a theoretical standpoint, it is interesting that SR offending and convictions had such a low prevalence in adulthood that analytically it was best to focus on the juvenile period. This finding is in accordance with Moffitt (1993), who argued that only a small proportion of people continue criminal behaviour through adulthood (especially the life-course persistent or LCP offenders; see Jolliffe et al., 2017).

In line with previous research (e.g. Farrington, 1989; Farrington et al., 2013a; West & Farrington, 1977), the prevalence of SR offenders was greater than official offenders, and the prevalence of SR offences was much greater than official offences. At younger ages (10–14 and 15–18) there were 2.8 self-reported offenders per conviction and at older ages (27–32 and 42–47) there were still 1.5. Similarly, at younger ages there were 23.9 SR offences committed for every conviction and at older ages there were 15.6. In both age ranges, offenders who committed more serious offences (burglary or vehicle theft) were generally more likely to be convicted, and more serious offences were more likely to result in a conviction. Also, those who committed fewer SR offences were less likely to be convicted, while almost all of those (about 88%) who self-reported 30 or more offences were convicted. At the most basic level, this suggests that successful SR offenders are those who avoid committing serious offences and who commit relatively few offences.

There are a number of possible reasons why the low seriousness of offending is related to successful SR offending. First, many serious offences provide more opportunity for evidence that will make the identification of the suspect more likely. For example, a burglar may leave behind forensic evidence (e.g. DNA) that would be less likely for less serious offences such as vandalism. Also, police and criminal justice officials prioritise the identification and prosecution of more serious offences, both to reassure the public and also because of the belief that those who commit more serious offences are more likely to repeat these offences and possibly escalate to even more serious offences.

Less serious offences may be less likely to result in a conviction because these are viewed with less priority by victims (i.e. these are less likely to be reported), the police (i.e. these are less likely to be investigated), and criminal justice officials (i.e. these are less likely to be prosecuted). Also, much previous criminological research has clearly demonstrated that minor offending perpetrated by young people is so common as to be normative (Moffitt, 1993), which may mean that it could not possibly all be investigated. More importantly, however, this same research shows that most young people naturally stop offending by their mid to late 20s. Therefore, a rational crime control approach may involve not pursuing minor offenders and offences, especially as it is also widely known that the official processing of low-level offenders can result in an increase in later reoffending (e.g. Farrington & Murray, 2014; Petrosino et al., 2019).

Regarding the relationship between successful SR delinquents and the number of SR offences, or the finding that an increased number of SR offences increases the likelihood of a conviction, this makes intuitive sense based on simple probabilities. This has been noted in other research based on prospective longitudinal studies (e.g. Farrington et al., 2003) and is an important indication of the predictive validity of self-reports of offending in regard to the ability to predict the later convictions (e.g. Farrington, 1973; Jolliffe & Farrington, 2014).

Childhood psychosocial predictors

A surprisingly large number of childhood factors measured at ages 8–10 (16) were found to predict later successful SR delinquents. The strongest relationships were for behavioural (e.g. low troublesomeness), and individual factors (e.g. high verbal IQ; low daring), but family (e.g. high parental supervision), parental (e.g. not having a convicted parent) and socio-economic (e.g. low school delinquency) were also significantly predictive. Many of these factors are similar to already established protective factors for offending (e.g. Farrington et al., 2016; Jolliffe et al., 2016; Lösel & Farrington, 2012). Also, many of these childhood factors predicted committing minor (as opposed to serious) SR offences, which was one of the key factors which may, in part, explain successful SR offending. For example, those with low troublesomeness and high verbal IQ were more likely to be successful SR delinquents and also more likely to commit minor SR offences.

The factors that were found to independently predict successful SR delinquents were having a low number of SR offences and restricting oneself to minor offences, as well as two individual factors (high school attainment; low daring) and two family-related factors (not having a convicted parent; not coming from a disrupted family). It is of note that high school attainment was predictive of being a successful SR delinquent. This is in line with past research on protective factors for offending by Jolliffe et al. (2016) using the Pittsburgh Youth Study. High school attainment may be associated with successful SR offending because it reflects high intelligence. That is, these individuals may have superior cognitive abilities, which they use to plan and execute their offences and minimise their likelihood of being detected. This could include committing an offence outside of the view of CCTV, for example, or even intimidating victims to decrease the likelihood of an offence being reported. However, even if these offenders are detected by police or prosecutors, their elevated cognitive abilities may facilitate success by allowing them to formulate good alibis or otherwise manipulate their way to avoid getting a conviction.

Low daring could facilitate success similarly, by increasing the care, thought and calculation in selecting victims or criminal opportunities. This is in line with both Chaiken and Chaiken (1985) and Youngs and Canter (2014), where engaging in more planned and instrumental criminal offending was associated with success. Being daring can be seen as a measure of low self-control, and low self-control is associated with convictions (e.g. Aguilar-Carceles & Farrington, 2017). It is therefore not surprising that high self-control (not taking many risks) should reduce the likelihood of criminal justice involvement.

The two family factors that were associated with successful SR delinquents were not having a convicted parent and not coming from a disrupted family. It is well established that convictions of parents predict convictions of their offspring (e.g. Besemer et al., 2017; Farrington et al., 2009a, 2017). It is possible that having a parent with a conviction might bias the official response to a young person’s behaviour. A victim, the police and/or prosecutors may be more likely to push for more serious consequences if the family includes ‘known’ offenders. It could also be that parents who have been convicted, having themselves had a negative experience of the criminal justice system, adopt an adversarial approach with the authorities who are dealing with their children, thus increasing the likelihood of their children being officially processed and convicted.

The significant inverse relationship identified between successful SR delinquents and a disrupted family may also reflect official bias. Police and other criminal justice authorities may view single parents as less able than traditional families to be able to provide the supervision and support needed to control a young person who is committing offences. It is also possible that single parents may support official action against their children because they do not feel that they can cope with them, and that a conviction will trigger statutory support or be a ‘wake-up call’ for the young person.

Limitations and suggestions for future research

This study, like all research, has limitations. First, the CSDD sample is predominantly urban lower class white British males, which may limit generalizability to other demographic categories. To what extent the same risk factors would be found for successful delinquents in other samples (e.g. female, middle class, rural, non-British, born more recently) are empirical questions for future researchers. As shown by Gilman et al. (2014), ethnicity is related to whether someone is charged or not, so research comparing different ethnicities’ success in SR delinquency is needed.

Second, in the current study, it was only possible to establish whether individuals had received a conviction or not, not where in the criminal justice system they may have been diverted. Therefore, some individuals may have been successful because they avoided detection, while others may have been successful because they influenced the criminal justice response after identification. Future research should examine the extent to which these different outcomes are predicted by the different childhood factors.

Third, this article essentially provides information about risk factors for successful SR delinquents who commit property offences. The extent to which the same risk factors would be found for successful SR delinquents committing other types of offences (e.g. violence or drug offences) are empirical questions for future researchers.

Conclusions and implications

This is the first study to investigate the childhood psychosocial predictors of successful SR delinquents. Remarkably, 16 predictors were found, in six categories (attainment, self-control, socio-economic, parental, family and behaviour). The most important independent predictors were high school attainment, unconvicted parents, low risk-taking and unseparated families. As a result, interventions should target these factors to decrease the likelihood of criminal justice involvement (e.g. Farrington & Welsh, 2007).

To the extent that some offending in adolescence is normative, it is desirable that it should be minor and infrequent and not lead to a conviction. Ideally, large scale early intervention programmes in childhood should be mounted. For example, interventions that increase academic attainment (e.g. the Perry pre-school intellectual enrichment programme, Schweinhart, 2013) might be useful. To support a child’s development as a low daring individual, the Stop Now and Plan (SNAP) programme might be useful (Augimeri et al., 2018; Farrington & Koegl, 2015). Also, it would be desirable to minimise family separations, which might be achieved by a family prevention programme such as multisystemic therapy, which significantly reduces family instability (e.g. Sawyer & Borduin, 2011). These kinds of early interventions would be beneficial both for children and for society in general.

Acknowledgements

We are very grateful to the Home Office, the Department of Health, the Department for Education, the Rayne Foundation, the Barrow Cadbury Trust, and the Smith-Richardson Foundation. For carrying out criminal record searches, we are very grateful to Gwen Gundry in the 1960s and 1970s, Lynda Morley in the 1980s, Sandra Lambert in the 1990s, Debbie Wilson in the 2000s, Owen Thomas in 2011–2012 and Lisa Robinson in 2017.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request and with appropriate safeguards.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work on this article was not funded.

Notes

1 The funding was insufficient to interview all males at ages 21 and 25.