More recent considerations of the age-crime pattern have focused on specific arguments emerging from the variance/invariance debate rather than forming generalised interpretations of its nature (as per the studies outlined previously).
Developmental criminology is one body of research that has grown in response to the variance/invariance debate despite Hirschi and Gottfredson’s (1983) perception of it as a problematic approach to the age-crime pattern.3 This approach is concerned with individual offending patterns across the life course (both within and between individuals), including age-graded risk factors, life events, and transitions, and whether their influences are immediate or enduring (Nagin, Farrington and Moffitt 1995: 111; Farrington 2003: 221). This means that life events such as (un)employment need not influence every offender’s criminal choices in the same way.
The majority of developmental criminology research has been conducted within a strictly criminological framework. However, two of the most influential works in the area are also sociologically informed. These are Sampson and Laub’s (1993) Crime in the Making: Pathways and Turning Points Through the Life Course, and its follow-up, Shared Beginnings, Divergent Lives: Delinquent Boys to Age 70 (Laub and Sampson 2003).4 Sampson and Laub’s objective was to investigate the under- researched role of age-specific transitions and trajectories as turning points for offence patterns. Therefore, by looking beyond sociology’s common focus on juvenile offenders (the core of Greenberg’s critiques of crime-related sociological perspectives), Sampson and Laub seek to determine whether criminal activity across the life course is simply a continuation from childhood to adulthood offending, or behaviour that emerges from an age-specific event (such as educational outcomes during childhood, or employment during adulthood).
3 See Gottfredson (2005) for a discussion of how developmental criminology has contributed to understandings of age effects.
4
For further discussion regarding the impact of marriage see Sampson and Laub (1990, 1992, 1997, 2003, 2005a, 2005b), Laub and Sampson (1993, 2001), and Laub, Nagin and Sampson (1998). Other empirical studies relating to developmental criminology that have been developed from the ideas of Hirschi and Gottfredson include Blumstein, Cohen and Farrington (1988), Nagin and Farrington (1992a, 1992b), Nagin, Farrington and Moffit (1995), Nagin and Land (1993), and Bartusch et al. (1997). For further reading on theoretical perspectives, see Moffitt (1993, 1997), Thornberry (1997), and Farrington (2003).
These objectives are explored within a framework that combines the principles of social control theory – namely Coleman’s (1988) conceptualisation of social capital – with the life course perspectives of Elder (1974, 1975, 1985 and 1992). They describe this framework as one which ‘incorporate[s] both stability and change over the life-course … [portraying] the importance of informal social ties and bonds to society at all ages’ (Sampson and Laub 1993: 7, 17). Social ties, conceptualised as a mechanism for ‘creat[ing] interdependent systems of obligation and restraint that impose significant costs for translating criminal propensities into action’ (1993: 141), are closely associated with age-graded social control institutions such as marriage and work. If an individual does not have a high level of social investment or social capital in such an institution, they are more likely to offend. By illustration, the life event of unemployment creates a distance between the individual and society. An unemployed individual can therefore be expected to have a greater propensity to offend due to the weakening of social ties and social control they may otherwise have developed through paid work.
Sampson and Laub’s earlier work (1993) reanalyses Glueck and Glueck’s (1968) groundbreaking three-stage longitudinal data collection, commencing in 1940, of the offending patterns of delinquent and non-delinquent American males at approximate ages 14, 25 and 32 years. They found that despite between-individual stability in offending patterns across demographic variables such as age, ethnicity, and socio- economic status, adult social bonds play a significant role in adult offending regardless of childhood criminal involvement.
Job stability is an example of this trend, with both delinquent and non-delinquent adult males indicating a greater likelihood of apprehension (particularly at age 25-32 years) if they experienced low job stability (the importance of this finding for this thesis will be explored further in the next chapter) (Sampson and Laub 1993). For delinquent males experiencing low job stability at age 17-25 years, 91 per cent were apprehended at the same age and 74 per cent were subsequently apprehended at age 25-32 years (suggesting that low job stability has an enduring effect on offending); it was also a greater predictor of apprehension frequency at age 25-32 years than 17-25 years. Job stability at 25-32 years was also found to be significant for the delinquent
group. At the respective ages of 25-32 and 32-45 years, 80 and 64 per cent of males with low-level job stability were arrested, compared with 44 and 48 per cent of those with medium-level job stability and 18 and 39 per cent with high-level job stability. In contrast, apprehension levels were somewhat lower at age 17-25 and 25-32 year olds if medium or high job stability were experienced at age 17-25 years; 62 and 47 per cent of males with medium job stability were apprehended at these respective ages, compared with 60 and 32 per cent of males with high job stability. These trends were generally consistent with those of the non-delinquent group. However, the strength of the association between unemployment and apprehension trends was twice as strong for this group at age 25-32 years, and five times stronger at age 32-45 years; the impact of experiencing low job stability at age 17-25 years was also similar on their immediate and future apprehension trends.
Sampson and Laub (1993: 147) conclude that job stability ‘significantly modifies trajectories of crime and deviance’ regardless of prior involvement in criminal activity. Therefore, informal social capital, and the associated creation (or otherwise) of social ties between adults and institutions such as the labour market, has a direct impact on adult offending patterns. This finding is consistent with Greenberg’s (1977, 1983) suggestion that social control is an integral element of the age-crime pattern, and appears to override Hirschi and Gottfredson’s (1983) suggestion that sociological theories cannot account for the association.
Subsequent life history interviews by Laub and Sampson (2003) with 52 members of the delinquent group from Glueck and Glueck’s (1968) data set, now aged approximately 70 years, similarly found adult social bonds to play a significant role in creating turning points for within-individual offending trajectories, while agency was found to play a greater role than previously acknowledged. More importantly, the qualitative data suggests that interpretations of the association between age and crime as either variant or invariant may be dependent on the type of data consulted; variance in peak age of offending and the age of desistance at the individual level, but similarities in age-specific offending patterns at the group level, were equally shown. This finding concurs with Farrington’s (1986) argument that individual age- crime curves are more likely to reflect variance than aggregate age-crime curves, as
well as Hirschi and Gottfredson’s (1983) argument that life course perspectives emphasise variations in age-specific trends that would have occurred regardless of life events.
The findings emerging across these two studies lead Sampson and Laub (2005a) to conclude that the age-crime pattern is variant in some respects (primarily at the individual level), but is essentially an invariant association as Hirschi and Gottfredson (1983) promoted. Turning points play a significant role in shaping the criminal choices of individuals, and indicate some age-differentiation (as evident in their 1993 analysis of job stability), but generally influence individuals in the same way regardless of criminal history, demographic attributes, and intervening factors. Thus, ‘persistent offending and desistance from crime can be explained by a general age-graded theory of informal social control, that emphasizes social ties, routine activities and human agency’ (Sampson and Laub 2005a: 41).
More recently, O’Brien and Stockard (2009) examine Hirschi and Gottfredson’s (1983) invariant thesis via an assessment of the role of cohort replacement in the rise in young homicide offenders in the United States between the late-1980s and mid- 1990s. They explore this question using an estimable function approach, calculating the contribution of age, period, and cohort effects to variance in the age-crime curve of homicide arrest rates. They find that each of these effects makes a statistically significant contribution to variation in the age-crime curve for homicide. Age effects, however, make a much greater contribution than period or cohort effects (83 per cent, compared to 12 per cent and four per cent respectively). This finding indicates that Hirschi and Gottfredson (1983) were correct in arguing that the age-crime pattern is invariant, although the apparent influence of period and cohort effects also suggests that it is not ‘rigidly invariant’.
This conclusion is further supported by the powerful interaction between cohort replacement and period that emerges when the three effects are considered within a combined age-period-cohort model (O’Brien and Stockard 2009). Seventy-seven per cent of the difference between the observed age-period-specific homicide rates and the standard age-crime curve is attributed to cohort replacement. At the height of
homicide offences accounted for by 15-24 year olds in the United States in 1990, the age-period-cohort model accounts for all variation between the standard and observed age-crime curves. Subsequently, in 1995 the effects account for 60 per cent of the difference for 15-19 year olds, and 74 per cent of the difference for 20-24 year olds. Over both of these periods, observed rates for 15-19 year olds were seen to deviate from the standard age-crime curve more than the observed rates of 20-24 year olds. Nevertheless, cohort replacement explains only a little less than half of the gap between observed and standard age-crime curves for 1990 and 1995, which strongly suggests that the domination of the crack cocaine market at this time was also a significant (period) factor in the ‘epidemic of youth homicide’. By 2005, O’Brien and Stockard (2009) note a ‘mini’ cohort effect. This is evident in the departure of the age-specific arrest rates for 25-29 and 30-34 year olds from the standard age-crime curve (higher than anticipated rates for the individuals who would have been young during the period of rising young homicide rates). With regard to specific cohort effects, a one per cent change in relative cohort size and the percentage of non-marital births is concomitant with a respective 1.11 and 1.34 per cent change in the age-period-specific homicide rates.
The findings from the age-period-cohort model lead O’Brien and Stockard (2009) to conclude that cohort replacement is a significant source of variation for the age distribution of homicide offenders. Although this would appear to support Greenberg’s (1983, 1985) suggestion that cohorts are a potential source of variance to the age-crime pattern, O’Brien and Stockard argue that the association between age and crime pattern is invariant. This is because the age-crime curve is seen to be relatively constant over time once the influence of these effects has been accounted for. Again, therefore, this demonstrates how the nature of the age-crime pattern is open to interpretation; indeed, in an earlier analysis, O’Brien, Stockard and Isaacson (1999, discussed in greater detail in Chapter 4) concluded that the impact of cohort characteristics on homicide rates could be interpreted as either supporting or negating Hirschi and Gottfredson’s (1983) invariance thesis.