The first explanation is that prior participation in criminal behavior has a genuine behavioral impact that increases the likelihood of future participation. The second is that stable, unmeasured differences exist in criminal potential across the population. Many general theories of crime can be interpreted as suggesting one of these interpretations. This analysis used a panel data set assembled by Farrington and West in a prospective longitudinal survey of males from a working-class area of London. The data collection began in 1961-62, when most of the boys were about age 8. To the extent possible, all were interviewed at intervals of about 2-3 years until they were age 32. Parents, teachers, and friends were also interviewed, and information from official records was collected. The results suggested that the positive association is largely attributable to stable, unmeasured individual differences. Tables and 45 references (Author abstract modified)