Recidivism and the Availability of Health Care Organizations

Abstract

Incarceration has been identified as a cause of poor health in current and formerly incarcerated individuals. Given the high likelihood of being in poor health when exiting prison, it is plausible that health impacts recidivism. Furthermore, ex-prisoners cluster in disadvantaged neighborhoods that are unlikely to have decent health services. Currently, there is insufficient research to examine this relationship at an ecological level. In this study, we investigate the relationship between the availability of health care organizations (HCOs) and their changes over time with neighborhood level recidivism, and how these relationships may be moderated by neighborhood disadvantage. We determine that the effect of HCOs on recidivism is indeed moderated through disadvantage: as disadvantage increases, the negative effect of losing significant amounts of HCOs on recidivism accelerates. Our results suggest that while increasing HCOs in disadvantaged neighborhoods is important, keeping HCOs in place is equally important for moderating negative neighborhood level outcomes.

Keywords:

• recidivism
• health care organizations
• neighborhoods

Acknowledgments

We would like to thank the PSN Taskforce, Christopher Berk, Tracey Meares, Jeffery Fagan, Sean I. Lee, and Matthew Larson lending their expertise. This project was supported by Grant #2006-GP-CX-0,017 awarded by the Bureau of Justice Assistance through the Illinois Criminal Justice Information Authority. The Bureau of Justice Assistance is a component of the Office of Justice Programs, which also includes the Bureau of Justice Statistics, the National Institute of Justice, the Office of Juvenile Justice and Delinquency Prevention, and the Office for Victims of Crime. Points of view or opinions contained within this document are those of the author and do not represent the official positions of the US Department of Justice, or the Illinois Criminal Justice Information Authority.

Notes

1. Research shows that not only does incarceration negatively effects health, while in prison, unhealthy prisoners often do not use the health care available to them (Mallik-Kane & Visher, 2008); additionally, the health care they do receive is not necessarily adequate (MacReady, 2009; Restum, 2005; Visher & Mallik-Kane, 2011).

2. While there is debate about the organizational density of disadvantaged neighborhoods (see Small, 2008), these neighborhoods are still at a disadvantage for acquiring health services.

3. A neighborhood unit of a zip is far too large both spatially and population wise to be considered an appropriate neighborhood unit when modeling social phenomena. However, outside of the limitations of the data, there are good reasons to use zip codes as neighborhood units. When working with variables related to economic forces, such as organizational resources, a census block or block group is too small to capture the economic forces related to organizational resource availability. Small and McDermott (2006) make the following case for using zip codes: “the use of zip data as opposed to tract data may be increasingly appropriate as we are concerned with economic well-being and increasingly inappropriate as we are concerned with social well-being” (p. 1702). Given that this study explores organizational resources and changes in their quantity over time, a zip code is an appropriate neighborhood unit.

4. In this study, we consider reoffending to be both new crimes and technical violations of parole.

5. When including the percent black and percent unemployed in the factor, the loadings are such: percent black (.93), percent unemployed (.30), percent female-headed household (.94), percent on public assistance (.97), and percent in poverty (.90). The low factor loading for percent unemployed lead me to exclude it from the overall concentrated disadvantage factor. Also, concentrated disadvantage can be tied to nonblack neighborhoods; for this reason, percent black was also removed from the concentrated disadvantage factor. Additionally, the low factor loadings for concentrated disadvantage score relative to other ecological studies in Chicago may be due to the unit of analysis; the underlying concept of concentrated disadvantage may not precisely fit onto the zip code as compared other measures of “neighborhood” such as census tracts of neighborhood clusters. In this case, zip codes overlap some of the historically concentrated black high-poverty areas as well as the of the Latino high-poverty areas—this is especially true on west side of the city. It is still true, however, that overall levels of contracted poverty are higher among blacks than Latinos or whites, no doubt an artifact of the long history of segregation and high-rise public housing in Chicago. Therefore, the percent black and percent unemployed were included in the model separately.

6. A concern when creating the health care organization loss variable is a floor effect. In this case, for neighborhoods with very low numbers of health resources (<3) may not have health care organizations to lose. To be certain this was not the case, we carefully examined the total number of health care organizations by neighborhood over the years. The minimum number of health care organizations any neighborhood has is four; therefore, a loss of three or more health care organizations is always possible in the data.

7. A concern here is that the relationship between losses of health care organizations, concentrated disadvantage, and recidivism is not predicated on a certain threshold of the number of HCOs lost, but rather the percent lost between years and how that may interact with the baseline number of HCOs in a neighborhood. Put another way, is losing three HCOs in a neighborhood where only four existed the previous year (a 75% loss) the same as losing three HCOs in a neighborhood that had 100 HCOs the previous year? To be sure our finding was robust, rather than using a threshold of either loss or gain (i.e. gaining or losing three or more HCOs), we created a percent change variable ((current HCO—previous years HCOs)/ previous years HCOs). Next, we created a variable which split the sample in half based on the number of HCOs the neighborhood has is either above or below the median number of HCOs. Lastly, we interacted percent change with concentrated disadvantage. After creating these three variables, we ran three models to test the effect of the interaction between percent change and concentrated disadvantage on recidivism: (1) a model which replaces HCO loss with percent change and includes the interaction between percent change and concentrated disadvantage on the full sample, (2) the same model as above but on the sample of neighborhoods with numbers of HCO below the median, and finally, (3) the same model as above but on the sample of neighborhoods with numbers of HCO above the median. In all models, we see that there is a significant interaction with percent change and concentrated disadvantage; however, what is unclear from the results is how the effect of this interaction on recidivism is different across the three samples (full, low HCO, and high HCO) we employed. To explore this, we plotted the predicted probabilities for each of the models using low, medium, and high levels of concentrated disadvantage. What we see is the same pattern across all models: in neighborhoods with high levels of concentrated disadvantage, losing HCOs across years has a more drastic impact on recidivism, than areas with lower levels of concentrated disadvantage. As a result, we feel confident that our model with the two threshold variables adequately captures the issue at hand.

Additional information

Notes on contributors

Danielle Wallace

Danielle Wallace is currently an assistant professor at Arizona State University in the School of Criminology and Criminal Justice. She received her PhD in sociology from the University of Chicago in 2009. Her research interests include neighborhoods and crime, theories of disorder, and offender reentry and recidivism. She is currently heading the Fast Track Abatement Program Evaluation Project—a photo documentation and systematic social observation project that constructs the time line surrounding disorder, home, and neighborhood deterioration as abandoned buildings become slated for demolition.

Andrew V. Papachristos

Andrew V. Papachristos is an associate professor in the Department of Sociology at Yale. His research focuses on urban neighborhoods, social networks, street gangs, violent crime, and gun violence. His current research uses social network analysis to study how violence diffuses among populations of youth.