Delinquent activity and age of onset

View the most recent version.

One of the common findings of criminal careers research is that children who begin early to engage in antisocial behaviour and delinquency are more likely to become chronic and serious offenders. This implies that these "early onset" offenders would have longer careers incorporating more criminal activity (Farrington, 1986; 1992; Farrington et al., 2003; Piquero et al., 2007). The section on the duration of delinquent careers (above) established that the early onset offenders in these two cohorts do have longer delinquent careers, within the limits of the period of observation.

Figure 32 shows the relationship between the number of incidents in the career and the offender's age of recorded onset. In the 1987 birth cohort, there is a clear downward trend¹ in recorded delinquent activity with increasing age of onset: from an average of 5.4 recorded incidents in the career for those whose first incident was at the age of 8 to only 1.4 incidents for those whose first recorded incident was at 17 years of age. In the 1990 birth cohort, there is a similar pattern for those with ages of onset of 8 to 14, although the numbers of incidents are lower because members of this cohort are not tracked through the high-activity ages of 15 to 17.² However, contrary to expectations, children with very early onset (5 to 7 years of age) are not the most active offenders: on average, they are involved in no more recorded incidents during the period of observation than those with onset at 8 to 10 years of age. Since their delinquent careers last longer (see "The duration of the delinquent career"), this implies that their annual rate of recorded offending, averaged over the length of the career, is lower.

The downward trend in activity with age of onset from ages 8 to 17 (Figure 32) is not due to different periods of time "at risk" of committing crime: offenders born in 1987 whose first recorded incident was at the age of 17 were at risk of committing recorded crimes for the same 10 years as those whose first incident was at the age of 8: the fact that the first (recorded) incident of the former was very late in the period of observation simply reflects a lack of (recorded) criminal activity during most of the period.³ Nor does the downward trend necessarily reflect a higher propensity to commit (recorded) crime on the part of those with early onset. An inverse relationship between age of onset and the number of incidents in the career is also observed in simulated data in which the propensity to offend is the same for all persons.⁴ Further research is needed to determine to what extent this downward trend is due to systematic versus random differences in the propensity to commit (recorded) crime.

Similar but not identical results are obtained by plotting the percentage of chronic offenders (those who have five or more recorded incidents in their careers) against the age of onset (Figure 33). In the 1987 birth cohort, the percentage of chronic offenders is constant for ages of onset of 8 to 11 years, then decreases with increasing age of onset. The decrease in chronic offenders with increasing age of onset is also apparent in the 1990 cohort, but not for offenders whose recorded careers began before the 8^th birthday. Thus, both analyses indicate that within the age ranges covered by this study, very early onset offenders do not commit more crime and are not more likely to be chronic offenders than those with onset between the ages of 8 and 11. This finding appears to contradict the results generally reported in the literature of greater criminal activity by earlier onset offenders. However, there are a couple of possible explanations for this unexpected finding which are not necessarily inconsistent with the results generally reported in the literature. One possibility is that the much lower rates of offending which characterize the years from the 5^th to the 8^th birthdays⁵ offset the impact of early onset on the offender's total activity up to the age of 14: perhaps if these offenders were followed for a longer period, accumulated offending by those beginning their careers before the 8^th birthday might outstrip that of later-onset offenders. Another possible explanation is that the true rate of offending by 5 to 7 year olds may be higher than that of 8 to 10 year olds, but this may be hidden by a greater tendency on the part of the public to report, and of the police to record, the involvement of older children in criminal activity than that of younger children.

A rather different view of the relationship between the age of onset and chronic offending can be seen in Figure 34, which shows the percentage of all chronic offenders in the cohort accounted for by each age-of-onset group. It is the later-onset groups which contribute the greatest numbers of chronic offenders in these two cohorts: those in the 1990 cohort whose first incidents were at 12 and 13, and those in the 1987 cohort whose ages of onset were 12 to 15.⁶  The reason for this is that the number of cohort members with later ages of onset is so much greater than those with early onset that they contribute greater numbers of chronic offenders, although the probability of any one of them becoming a chronic offender is lower than for early onset offenders.

This effect can also be seen in Figure 35, which shows the total recorded criminal activity (number of incidents) for each age-of-onset group. In both cohorts, the early onset offenders as a group are responsible for very little volume of crime compared to those who began offending in their early teens. It appears from the pattern for the 1987 cohort that offenders whose first recorded incident took place at the age of 14 were, as a group, responsible for the greatest volume of crime, followed by those with onset at 15 and 13, then by those with onset at 16, 17, and 12. However, the numbers for those with onset at 16 and 17, are probably underestimated, due to data truncation.⁷

This does not necessarily contradict other studies which have found that early onset offenders are responsible for greater amounts of crime. "Early" and "late" are relative terms. Most studies of delinquent and criminal careers using police or court data have necessarily begun tracking offenders at the 10^th birthday or later. Thus, onset at 13 to 14 years old, which is relatively "late" in the present study, would be relatively "early" in a study which tracked offenders from 10 to 32 (Farrington, 1992) or from 12 to 21 years of age (Carrington et al., 2005). Also, it may be, as writers following Moffitt (1993) would predict, that the early-onset offenders in this study whose careers do continue into adulthood will eventually be responsible, as a group, for more recorded crime than the later-onset offenders, because they will continue to offend for considerably longer and/or at a higher rate. However, this seems unlikely, given the much greater number of later-onset offenders. Avoiding relative and ambiguous terms such as "early-" and "late-" onset, and predictions of the future, what can be said with certainty is that although the childhood-onset offenders⁸ in the present study committed more recorded crime per offender than the adolescent-onset offenders (Figure 32),⁹ they were responsible as a group for much less recorded crime during the period of observation (Figure 35), because there were so few of them: they make up only 11% of all offenders in the study and were responsible for only 17% of the total volume of recorded crime attributed to members of these two birth cohorts.

---

Notes

1. Apart from an anomalous drop for offenders whose first incident was at the age of 9.
2. See "Age and the rate of offending".
3. The total frequency of recorded offending of offenders whose careers began during the last two years of observation (2004 and 2005, indicated by dotted lines in Figures 32 and 33) may be underestimated because of data truncation: that is, because their careers continue into the unobserved period after 2005 (see "The duration of the delinquent career"). These are offenders whose careers began at the ages of 16 and 17 in the 1987 birth cohort, and 13 and 14 in the 1990 cohort. While such underestimation may also apply in the case of other offenders, it is unlikely to be related to the age of onset, and is therefore unlikely to bias the analyses reported here, as the analysis of the duration of careers (above) found that, with the exception of careers beginning in the last two years of the period of observation, late-onset offenders were probably no more likely than early-onset offenders to have careers which continued past the end of the observation period. There is no doubt that in restricting analysis to activity up to a certain age, this study does not provide a true estimate of the lifetime delinquent and criminal activity of these offenders - nor does it attempt to - but the impact of the restricted period of observation should be approximately the same, regardless of the age of onset (except for the last 2 years).
4. I thank Paul Verbrugge for pointing this out.
5. See "Age and the rate of offending".
6. As with the earlier analyses, the percentages for offenders with onset in the last 2 years of the observation period are probably depressed by the truncation of the data. Therefore, the increase in percentages up to the age of onset of 14 in the 1987 birth cohort is probably more meaningful than the decreases at 13 and 14 in the 1990 cohort. Similarly, the decreases for those with ages of onset of 16 and 17 are probably exaggerated by data truncation.
7. See related notes on earlier analyses.
8. With recorded onset before the 12^th birthday.
9. Whose recorded onset occurred on or after the 12^th birthday.