Abstract
The vast majority of tests of major criminological theories have been conducted on youths in school settings. Following Hagan and McCarthy’s (1997) admonition to engage in “street criminology,” the current study surveyed 254 juveniles drawn mainly from impoverished neighborhoods in Uruguay, who were either confined or in an alternative justice program. Due to limited literacy skills, all youths were interviewed regarding theoretical measures and their delinquent involvement. The results revealed that self-control was unrelated to delinquency in this sample. By contrast, social learning factors and stressful life events were found to increase delinquent participation. The results suggest that the effects of the field’s core theories are general, accounting for as much or more variation in the sample of street youths as among youths attending conventional U.S. schools. The policy implications of the findings are explored.
FUNDING
This research was funded by the Sectorial Commission of Scientific Research (CSIC in Spanish), Universidad de la República, Uruguay.
Notes
1. The possibility of escaping from an institutional facility is only available to incarcerated adolescents. However, escaping is a phenomenon that reflects a different dimension of delinquent involvement. The number of escapes appears as relevant for those adolescents with a higher level of involvement in delinquency, and allows one to distinguish different incarcerated behaviors among them. Youth incarcerated for longer periods and more often may show lower levels of involvement in crime due to longer incapacitation periods. Thus, we decided to include a rule violation that captures a dimension of behavior relevant within the periods they are confined. Alternative estimations of every model were performed using a different delinquency index in which the number of escapes was excluded from the factor analysis. The only difference between these estimations is the significance of informal and police labeling in the multitheory models. Also, differential reinforcement is not significant on these models. All other results hold in the OLS, robust OLS, and WLS estimations.
2. The use of dichotomous items on an offender population raised specific concerns in terms of the variability of the self-control scale. However, its standard deviation is .18 and the histogram shows an acceptable degree of variability across the scale.
3. Due to the fact that the data was MCAR, an alternative course of action would have been to delete the cases with missing values. However, to be consistent with the approach taken with the quantitative variables, in which data were imputed, and to maximize the information used for the estimations, we decided to replace the missing values with the variable modes. Given that the data is MCAR, the mode of all variables is close to the mean, and the amount of missing data is small (6% and 1%), this strategy should not introduce bias into the estimations.