a study of criminal careers makes at least three demands of the data on which it is based. They must:

• concern individuals

• concern a period of time

• inform about the time of the incidents

Data for such a longitudinal study can be gathered either through surveys of self-reported criminality or through person-based registers.

Self-reported crime data have the advantage of including a wider spectrum of offenses committed by individuals than that generally found in official registries. Unlike register-based research, undetected crimes and those not reported to the police are measured in studies of self-reported criminality.

The gathering of data through self-report also has some disadvantages, however. The principal one – especially for longitudinal studies – is selection bias. There will be some who do not wish to participate in the study at all, and this refusal rate will increase with repeated interviews. Even if attrition analyses can show that dropouts do not differ from the study sample in terms of basic demographic characteristics, there may still be reason to be skeptical, especially if the study includes individuals with great social and criminal strain. It can be very difficult to establish – not to mention maintain – contact with those who commit a lot of crime, and who often live very vulnerable lives. Furthermore, it is questionable whether adults, and especially high-rate offenders, will be as honest in their reporting as children and young people (Mathur, Dodder, and Sandhu, 1992).