ABSTRACT
This study investigates the impact on recidivism of offenders’ misconduct in prison, net of offenders’ socio-economic and criminal backgrounds, longer imprisonments, and of other, unobserved factors explaining selection into prison. The data come from representative samples of offenders (N = 4524) and prisoners (N = 1848) convicted of at least one crime in Catalonia, Spain, during a 5-year period (2010–2015). We applied a system of four simultaneous equations to estimate which factors increase the most the risks of reoffending, focusing on the role of prison misconduct, and using better methods than previously to correct for selectivity into prison. We also identify the characteristics of the offenders most likely to misbehave and reoffend – chronic offenders with complex criminal histories. Our results confirm that misconduct significantly predicts reoffending. It is not just a reflection of offender’s problematic dispositions imported into prison, but also a sign of maladjustment to imprisonment that undermines its possibly deterrent or capacitating effects.
Acknowledgments
We are grateful to the Secretaria de Mesures Penals, Reinserció i Atenció a les Víctimes of the Departament de Justícia of the Generalitat de Catalunya for giving us access to their data registries and for their continuous support for this research. We also thank Elena Larrauri for her useful comments on previous versions of the paper.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. If there were more than one crime committed by the same person on the same date, the most severe was considered.
2. This random sample of offenders was drawn as part of a larger study in which data on the entire population of offenders convicted for an Intimate Partner Violence crime (IPV) was collected for the same period (N = 6,984). In the analyses, we pool the two datasets (the sample of general offenders and the full cohort of IPV offenders), eliminating any IPV duplicates, and give IPV offenders their real (much smaller) weight in the sample. We use Huber-Sandwich robust standard errors to correct the potential bias introduced in the probability tests by the weighting procedure.
3. All the statistics provided hereafter are weighted, i.e., they all reduce the number of IPV offenders to their true number in the sample of all offenders.
4. To build it, we used all offenders incarcerated for the selection crime, whether we observed their release.
5. = 0.37.
6. = 0.43.
7. Since we have a large sample size and many covariates, the results are similar to those we would have obtained with propensity score matching.
Additional information
Funding
Notes on contributors
Jorge Rodríguez-Menés
Jorge Rodríguez-Menés is a Serra Hunter associate professor of Sociology and Criminology at the Universitat Pompeu Fabra of Barcelona (Spain). He has a PhD in Sociology from Northwestern University (USA) and worked as a lecturer and senior lecturer in Sociology and Quantitative Methods at the Universities of Reading and Kent (UK).
Amalia Gómez-Casillas
Amalia Gómez-Casillas holds a PhD in Demography and works at Centre for Demographic Studies (CED-CERCA). She provides technical support to the European Institute for Gender Equality (EIGE) as an External Expert in Police and Judicial indicators of gender-based violence. She has been a lecturer in Criminology at the Universitat Pmopeu Fabra of Barcelona.
Fernando Ruíz-Vallejo
Fernando Ruíz-Vallejo Assistant professor at the Institute of Public Health of Javeriana University (Pontificia Universidad Javeriana) in Bogota (Colombia). He has a doctorate in demography, a master’s degree in population and development, and a degree in sociology.