The longitudinal study Crime in the Modern City (CrimoC) analyzes developments of delinquent and deviant behavior, their (mainly social) causes as well as the effects of preventive measures and control interventions from age 13 until age 30 (in 2019). An important difference to other studies, which are often cross-sectional, is the repeated survey of the same participants at fixed intervals in a so-called panel design. The information gained by this method makes it possible to analyze the differences between persons or groups (interindividual changes) as well as the individual developmental pathways (intraindividual changes).
The empirical results confirm the expected phenomenon of a ubiquitous prevalence of juvenile delinquency. 84 percent of boys and 69 percent of girls reported having committed an offense at least once between ages 13 and 18 (all offenses excluding internet offenses and drug use). With regard to violent offenses (including assault with and without a weapon, and robbery), the proportions were similarly high with 61 and 37 percent, respectively (Figure 1).
Figure 1: Lifetime prevalence age 13 to 18. Cumulated panel rates in percent. Duisburg 2001 to 2007; n=1.307.
The high rate of deviant behavior in early adolescence is followed by a clearly discernible process of so-called spontaneous remission, which underlines the appropriateness of diversion measures applied by juvenile justice in cases of first-time or occasional offenders. It is noteworthy that the highest rate of self-reported delinquency occurs much earlier than that of officially registered delinquency. In Duisburg, this was already the case at the beginning of adolescence (age 14-15), while in the police statistics the maximum occurs towards the end of adolescence. The decline of self-reported delinquency began in Duisburg at age 15 to 16. At age 17 the level of delinquency was already lower than at age 13. By age 18 it had decreased to half or even one third of the maxima at early adolescence (Figures 2 and 3).
Figure 2: Prevalence for self-reported theft and vandalism, age 13 to 22. Valid percentages, Duisburg 2002 to 2011.
Figure 3: Prevalence for self-reported violent offenses, age 13 to 22. Valid percentages, Duisburg 2002 to 2011.
Intensive offenders play a major role in criminological research as well as in policy debates. Typically (and fortunately), the amount of intensive offenders is very small. In Western countries they (up to now) made up for 5 to 7 percent of a juvenile age group. These offenders are, however, characterized by high rate of offense frequencies, in particular with respect to violent offenses. In our understanding the group of intensive offenders accounts for at least 50 percent of all offenses and at least 75 percent of all violent offenses (assaults with and without weapons, robbery offenses) being committed by all offenders of their age group. In Duisburg, at age 14 and 15, this group (classified by five and more violent offenses per year of age) made up for 6 percent of the entire sample (9 percent males, 3 percent girls) but reported half of all offenses and more than 80 percent of all violent offenses. The amount of intensive offenders is much smaller, in particular for girls, if serious violent offenses (assault with a weapon, robbery offenses) are considered only (3 and less than 1 percent, Figure 4). Especially in the 1980s and 1990s it was assumed that most intensive offenders remain delinquent well into adulthood, particularly those who showed deviant behavior in childhood. In the present study, however, also the proportion of intensive offenders decreases by age 16, and thus earlier than expected.
Figure 4: Prevalence of intensive offenders (five and more offenses in the past year), age 13 to 22. Self-reported delinquency. Valid percentages. Duisburg 2002 to 2011.
The development of delinquency within the population does not follow a homogeneous pattern. The (unobserved) heterogeneity of developmental patterns can be analyzed by means of Growth Mixture Models (GMM, here as Latent Class Growth Analysis, LCGA, one derivate of GMM). Using the average total incidence (frequency) rates of annually self-reported delinquency (offenses per respondent), the best model resulted in seven developmental groups (Figure 5):
Figure 5: Trajectories of self-reported delinquency. Seven wave panel data, age 13 to 19. Latent Class Growth Analysis (LCGA), negative binomial distribution, maximum one wave missing, observed data. Duisburg 2002 to 2009, n=1.895.
Three of the four largest groups consist of (almost) non-delinquents (43 percent), or of offenders whose delinquent behavior remains either at a very low level (14 percent), or starts at an early age, but declines quickly thereafter (low level declining, 9 percent). Members of the fourth largest group offended at an higher, yet still moderate rate, peaking at age 15 and decreasing also towards zero at the end of adolescence (15 percent adolescence limited offenders).
Particularly meaningful from a criminological and criminal-policy perspective are the three remaining groups of persisting intensive offenders(8 percent), early declining intensive offenders (6 percent) as well as late starters (6 percent). Among the most strongly affected persistent intensive offenders the incidence rate increased at an early stage and reached a maximum at age 16 of on average 44 offenses per person. Subsequently, it decreased remarkably so that at age 19 it was below the level of age 13.
Such a reduction of the intensity of delinquent behavior is not compatible with the assumption of a trajectory of life-course-persistent behavior. Furthermore, the common assumption of an early onset of delinquent behavior as one of the best predictors for future serious delinquency has to be put into perspective against the background of the following results: Nearly half of the adolescents, who showed the same high frequency rate of offending at age 14 as the persistent intensive offenders, started reducing their delinquent activity significantly over the course of the next year (early declining intensive offenders).
Furthermore, the development of late starters does not support the assumed prominent importance of an early onset. Approximately 6 percent of the respondents in Duisburg are classified as late starters (other studies report up to 15 percent). These offenders started offending at age 16 and their frequency level during the ages of 17 and 18 is considerably high – second only to the group of persistent intensive offenders. However, the level of offending of the late starters also decreased at age 19. These results appear to be quite similar to those from Anglo-American studies. Nevertheless, the declining tendencies seem to start earlier in Duisburg. Although the strong dynamics of delinquent behavior make its prediction difficult, without these dynamics preventive interventions could not be successful in providing a realistic chance for change.
To explain the different developmental phenomena a Structural Dynamic Model was developed (figure 6). This model integrates different criminological theories and distinguishes between indirect distal (social structure, social milieu) and direct proximate causes (delinquent peers, delinquent norm orientation, figure 6). Furthermore, the impact of control interventions by police and juvenile penal courts is taken into account.
Figure 6: Structural Dynamic Model for the Analysis of Delinquency.
During adolescence, analyses of the CrimoC-data allowed, in general, to distinguish two explanatory patterns: a pathway into violent offending, and a pathway into conformity. Juvenile violent offenders are especially influenced by belonging to delinquent peer groups in which delinquent norm orientations and delinquent behavior mutually reinforced the cohesion of the group. The significance of this violence-promoting cycle, however, decreases strongly with age. Preventive measures should particularly focus on the dynamics of delinquent peer groups, their value and norm orientations, as well as on alternative leisure time activities, for example, in the framework of social street work.
On the other hand, a pathway into conformity is based on stable social bonds to family, in particular parents, and school. Both play a decisive role in the learning of conventional norm orientations as well as socially adequate behavior in situations of conflict. Empathetic parenting as well as good relations between students and teachers were of particular importance. Bad relationships between students and teachers can promote delinquent behavior. Prevention work in school should first of all be conducted by teachers to whom the students relate most closely, and supported, if necessary, by external supervision.
As expected, the consumption of violent media does not lead to an increase of violent delinquency, but to a stronger approval of attitudes that support violent behavior. With respect to prevention, it is recommended to supervise the consumption of violent media and to impart alternative attitudes not approving violence.
In contrast to findings of several other studies, the migration background in Duisburg is criminologically rather insignificant. Youths of Turkish origin show stable ties with regard to family, school and neighborhood. The stronger approval of traditional and religious value orientations in this group, which is accompanied by less risky behavior in leisure time activities, also appears to inhibit delinquency. An important key in minimizing the risks of delinquency among youths with a migration background is the promotion of educational participation. The better the integration in the educational system, the less attractive are delinquency and violence.
Initial analyses on the transition from school into professional life indicate that stable conditions in vocational or academic training and a resulting smooth transition into the respective profession correspond with a decrease of delinquent behavior. Simultaneously, these transitions promote the emergence of conforming norm orientations. Even if intensive delinquent behavior can be observed towards the end of adolescence (especially among the persisters and late starters), measures that provide youths with a pathway into a stable vocational and professional environment are particularly helpful. In this context, delinquent behavior in early adolescence does not appear to exclude a successful transition into professional life.
Based on entries in the juvenile justice register (Bundeszentralregister and Erziehungsregister) between ages 14 and 20, we could analyze how interventions by police, prosecution and courts affect the development of delinquency. The results show that already low level control interventions (mainly diversion decisions) can lead to a stronger association with delinquent peer groups, which in turn promotes delinquent behavior. When reacting to juvenile delinquency, it may therefore still be recommendable to limit the penal measures to the fewest necessary.
For further results and analyses, see the following publications:
Boers, K.; Reinecke, J.; Mariotti, L. & Seddig, D. (2010): Explaining the Development of Adolescent Violent Delinquency. In: European Journal of Criminology 7, 499-520.
Reinecke, J. & Seddig D. (2011): Growth Mixture Models in Longitudinal Research. In: Advances in Statistical Analysis, 95 (4), 415-434.
Seddig, D. (2014): Peer Group Association, the Acceptance of Norms and Violent Behaviour: A Longitudinal Analysis of Reciprocal Effects. In: European Journal of Criminology 11 (3), 319-339.
Seddig, D. (2015): Crime Inhibiting, Interactional and Co-Developmental Patterns of School Bonds and the Acceptance of Legal Norms. In: Crime and Delinquency (in print).
Theimann, M. (2015): School As a Space of Socialization and Prevention. In: European Journal of Criminology (in print).