Patterns of Offending BehaviourAlthough criminologists often explore the occurrence of offences, the patterns of these offences are rarely examined. However, it is precisely these patterns of offences which could help to predict and gain a greater understanding of offending behaviour and its causes as well as contribute to the improvement of prevention. In this project, offence patterns were determined by looking for offence combinations within the criminal career of individuals. To this end, the research analysed criminal specialisation in several ways. Firstly, specialisation was explored within the offence categories and secondly, the relationship between different offence categories was analysed to define the occurrence of typical offence patterns within that specialisation. |
| Project category: | Doctoral dissertation |
| Organizational status: | Departmental project |
| Project time frame: | Project commences: 2003 Project ends: 2007 |
| Project status: | Completed |
| Project language(s): | German |
Head(s) of project:
The aim of the research project was to determine similarities between offence types on the basis of data from the Freiburg Cohort Study . The similarity of offences is determined empirically by the frequency of the common appearance of combinations of offences within the criminal careers of individuals.
It is assumed that offences committed by one and the same person show similarity. First, a criterion for the similarity of offences was created and fixed. Similarity can be defined by means of the penalty for the offence or by means of other criteria. In this project, none of these a priori categorisations were used. Instead, the similarity of offences was identified empirically. The frequency of offence pairs within the criminal careers of individuals were used as a similarity measurement. The frequency of offence pairs were compared with the frequency of random offence pairs. If one offence pair occurs more often than it would in accordance with a random distribution of offences within criminal careers, then these two offences are similar. Alternatively, if one offence pair occurs more seldom than anticipated, then these two offences are dissimilar. Therefore the Adjusted Standardized Residue (ASR) is suitable for measuring and analysing offence similarities. In contrast with other studies on the topic of specialisation, in the present project the transition from one offence to another is not important. All combinations of offences within the criminal career of a person are equally important, not just the next offence.
For this study police and court data differentiated by gender and by nationality was used.
One conclusion of the study is that all offences are similar with themselves. So specialisation is visible. The tendency towards specialisation is greatest for fraud, drug and sexual offences. The study also reveals a tendency for ‘specialisation’ regarding court registered traffic offences.
The similarity of offences is represented with the help of the Multidimensional Scaling (MDS) method. MDS is a method that represents measurements of similarity among pairs of objects (here offences) as distances between points of a low-dimensional multidimensional space, especially a two-dimensional space. MDS attempts to model data as distances among points in a geometric space. The main reason for using this approach is to achieve a graphical display of the structure of the data, one that is much easier to understand than an array of numbers. With MDS, the pattern of offences can be clearly depicted.
The following figure shows the offence pattern of court registered German males produced with MDS. What are important in this diagram are only the distances between the points in relation to each other: the factors of "above and below" or "left and right" are irrelevant. The closer the points are (short distance) the more similar the offences are. A distance of 0.7 units relates to an ASR of zero. That means offences with a distance less then 0.7 units are similar. If the distance is greater then they are dissimilar.
A two-dimensional MDS representation of the offences from court registrations (German men)
Similar offences are (the points have a small distance between each other) assault, aggravated assault, libel, violation of privacy, sexual crimes and homicide. The similarity of violent offences permits the conclusion that there are common causes and motives for these offences. Bodily injury caused by negligence and violent crime are not similar to one another. Drug offences are similar to theft and to fare dodging, which appears to suggest a relationship between drug offences and drug-related crime. The point for traffic offences is far away from the points for the other offences. Therefore the dissimilarity of traffic offences from ‘normal’, classical and conventional crimes is apparent. The quality of the two-dimensional presentation of the offences is very good (R² = 0.76).
In difference to court registered cases, police registered aggravated theft among German males, is located relatively far away from the other offences and is thus dissimilar to all other offences by police registered German males. Aggravated theft (burglary) often occurs in a series and is therefore often registered multiple times in the police data. Thereby the degree of specialisation is very high amongst crimes of burglary when registered with the police and accordingly the frequency of combinations from other offences with burglary is seldom. In court data a serial set of burglaries may come to a single decision in a single court case. A one-time registered burglary in court data could thus be registered multiple times in police data. Therefore the dissimilarity from aggravated theft with other offences which is apparent from the data of police registered males is not in fact apparent amongst court registered males. Offences against the Foreigners Act by foreign nationals seldom occur in combination with common crime. Only forgery and fare dodging reveal a relationship to ‘foreign’ crime. Special results were aroused by the analysis of separate age-groups. Many results were consistent in different age-groups, but a few results show that the relationship between different offences varies in accordance with different age phases. Especially amongst sexual offences, a tendency exists to similarity between sexual crime and violence by young people. This is not apparent with adult offenders. The study also concentrated on determining types of activity over the life course. In a first step, offence patterns of five-year periods were created with the help of probabilistic cluster analysis. In a second step, variations were analysed by age. The probabilistic cluster analysis pointed to similarities amongst violent offences. The analysis offered one cluster of violent offences. The estimated probability of an age strip in this cluster having a violent offence (assault, violation of privacy, robbery, homicide, sexual crime) and also libel and criminal damage is above average. Another result is a small cluster of versatile offences. All offences in this cluster are above average and the average number of offences in an age strip is very high. In this cluster every age strip contained on average six offences which is three times more then all age strips on average contained. Therefore this cluster contained ‘chronic’ offenders. The probabilistic cluster analysis produced two clusters for court registrations of traffic offenders. One cluster contained traffic offences without personal injury, the other with personal injury. An additional result of the probabilistic cluster analysis offered an insight into the routes taken by offenders in their criminal careers. Two issues of the criminal career – specialisation and offending pathways – were analysed. A specialist offender is defined as someone who stays within the same offending cluster. Specialisation increase as offenders grow older, without taking into account desisters. For any offending cluster a tendency of specialisation was visible, in the sense that the offenders stay in the same cluster in the five-year period before and in the five-year period after. Nevertheless the biggest group within the offending pathways is the group of offenders who desist from crime. Only in the cluster of the ‘chronic’ offenders is the before and after ‘desister’ group not the biggest.Publications (selection):
-
Tetal, C.: Analyse von Deliktsähnlichkeiten auf der Basis von Individualdaten der Freiburger Kohortenstudie.
Reports on Research in Criminology, Berlin 2008, 276 p.