Abstract
Empirical research shows that immigrants have lower rates of offending, arrest, and incarceration than the native-born. However, previous work has not examined whether this relationship extends to recidivism. The current study addresses this gap in the literature by comparing recidivism outcomes of 192,556 formerly incarcerated native- and foreign-born individuals released from Florida prisons. Using multiple analytic methods, including logistic regression, propensity score matching, and survival analysis, we find that immigrants are less likely to reoffend than their native-born peers. We conclude with a discussion of our study’s findings for future research and policy and practice.
Acknowledgements
The authors thank Carter Hay, Alex Widdowson, Jennifer Brown, and Brendan Lantz for their helpful comments and suggestions, as well as Sonja Siennick and Jillian Turanovic for their statistical advice. We would also like to thank the Florida Department of Corrections for permission to use their data. The views expressed here are those of the authors and do not reflect those of the Department of Corrections.
Disclosure Statement
No potential conflict of interest was reported by the authors.
Notes
1 Due to the large percentage missing for the marriage variable, all analyses were estimated prior to imputation as well. Additionally, all analyses were conducted with and without the marriage variable. Results were consistent regardless of whether the imputed data were used and regardless of whether the marriage variable was included.
2 Prior research suggests that criminal involvement may vary by generational status (Bersani & DiPietro, Citation2016; Morenoff & Astor, Citation2006; Rumbaut & Ewing, Citation2007; Sampson et al., Citation2005), and we would expect the same to be true for recidivism. However, as this is administrative data, our foreign-born measure is limited to country of birth, rather than a more nuanced measure based on generational status.
3 We also calculate and report t-statistics for the difference in means between the native-born and foreign-born samples, before and after matching. However, t-tests are substantially influenced by sample size, while standardized differences between groups are not (Austin, Citation2009). Therefore, given the large sample size here, we focus our attention on percent bias values when discussing whether the matching procedure achieved balance.
4 While the means for other race, age, and prior prison commitments are significantly different according to the t-test results, we remind the reader that this may be due to the large sample size (Austin, Citation2009). Therefore, we rely on percent bias statistics for assessing balance; given that the percent bias value for these three covariates is less than 20%, we accept these variables as balanced.
Additional information
Notes on contributors
Javier Ramos
Javier Ramos is a doctoral candidate in the College of Criminology and Criminal Justice at Florida State University. His research interests include immigration and crime, communities and crime, and the link between research and public policy.
Marin R. Wenger
Marin R. Wenger is an assistant professor in the College of Criminology and Criminal Justice at Florida State University. Her research interests include quantitative methods, communities and crime, stratification, and race/ethnicity.