This work is primarily aimed at identifying and measuring the relative explicatory ability of the following three hypothetical causal dimensions of criminal recidivism: a) socioeconomic variables, namely: marital status, children, place of birth, age, religion, race, educational level and qualification level of the occupation; b) type of offense, according to eight simplified types: homicide, assault and battery, larceny, robbery, embezzlement, possession and use of illegal drugs, and rape); and c) prison conditions, including the availability and qualification of prison personnel - both in security and management - prison work by inmates, professional training and general educational level, available spaces and overpopulation, and social programs for inmates. The work summarizes and evaluates international literature on criminal recidivism, both in methodological and in theoretical terms, presenting a recent historical overview on crime in the southern Brazilian state of Rio Grande do Sul, as the background for analyzing recidivism itself. Based on official longitudinal data from 1989 to 1997, and the state's general inmates registry, a series of regressions are calculated by the step-wise method, relating the three dimensions in question with the variable of recidivism, dichotomizing it. The socio-economic dimension explains 32% of recidivism, crime type explains other 9% of recidivism, and prison conditions explain an additional 16%, suggesting a wide margin of recidivism reduction by prison policies. Some closing considerations are presented in relation to the "new penology" and prisonal philosophy in Brazil.
criminal recidivism; socioeconomic base; offense; prison conditions; prison assistance programs
