The paper first presents an exemplification of the problems associated with, and the limited explanatory power of individual-level prediction for the prediction of crime at the aggregate level. Following that, the explanatory powers of individual-level variables at the macro level are tested. This involves a sociological analysis of psychological data, examining measures of intra-family conflict over time. Results indicate that the predictive power of individual childhood or teenage properties is too weak to explain total crime or specific types of crime. Changes in aggregate measures of conditions during upbringing are not easily compatible with changes in trends in crime; and other variables relating to changes in the opportunity structure and changes in the family upbringing can account for the development of crime trends. As an example, the trend in theft and violent crime in Sweden since World War II was shown to mirror the trends followed by the opportunity structure for goods liable to theft, in this case cars, and by levels of total alcohol consumption. Assumptions regarding changes in youth's level of motivation to commit crime, or their capacity for self-control, do not seem to be required to explain the increase in crime witnessed during this period. The data was collected from the Swedish level-of-living surveys. Tables, figures, appendix, and references