The Carceral State, Social Disorganization, and Firearm Homicides in Chicago

This cross-sectional study analyzes the association of race-specific incarceration with race-specific firearm violence rates in Chicago, Illinois, and whether social or familial instability mediate these associations.

Key Points

Question

What is the association of incarceration rates with the perpetuation of modern-day firearm violence and how does familial instability mediate this association?

Findings

In this cross-sectional study of 46 312 census blocks, Black-specific incarceration rates were found to be associated with Black-specific firearm homicides, but there was no association for Hispanic or White incarceration rates. In the association of Black incarceration rate with Black firearm homicide rates, social vulnerability did not mediate the association, while percentage of single-parent households mediated 23% of the association.

Meaning

These findings suggest that social disorganization resulting from mass incarceration may perpetuate urban firearm violence and worsen racial disparities therein.

Abstract

Importance

Black and other racially minoritized groups are overrepresented among those who experience firearm homicide. There has been a stark increase in incarcerated populations in the US since the 1980s, largely due to differential drug sentencing, of which racially minoritized individuals are also overrepresented; social disorganization theory postulates that community and family instability resulting from incarceration can further worsen crime.

Objective

To understand the association of race-specific incarceration with race-specific firearm violence rates in Chicago, Illinois, through the lens of social vulnerability and family instability.

Design, Setting, and Participants

This cross-sectional study with a retrospective cohort design utilized homicide data for the city of Chicago from January 1, 2001, to August 31, 2019. Demographic data at the census block level was obtained from the 2010 decennial census. Incarceration rates were obtained by race and ethnicity at the census tract level from the Opportunity Atlas. Data analysis occurred from January to June 2023.

Exposure

Race-specific incarceration rates were the primary exposures of interest. The Social Vulnerability Index and single-parent households were studied as mediators.

Main Outcomes and Measures

Race-specific firearm homicide rates were the outcomes of interest. Structural equation modeling was used to understand the mediating effect of social vulnerability and single-parent households on the association of incarceration with firearm homicides.

Results

A total of 46 312 census blocks were evaluated. Black-specific incarceration rates were found to be associated with Black-specific firearm homicides (incidence rate ratio [IRR], 1.70; 95% CI, 1.50-1.94), but there was no association for Hispanic incarceration rates (IRR, 0.98; 95% CI, 0.75-1.28) or White incarceration rates (IRR, 1.13; 95% CI, 0.39-1.16). In the association of Black incarceration rates with Black firearm homicide rates, social vulnerability did not mediate the interaction, but the percentage of single-parent households mediated 23% of the interaction.

Conclusions and Relevance

This study found that higher rates of incarceration were associated with increased rates of firearm homicides among Black communities alone; this association was found to be mediated partially through the density of single-parent households in these areas, suggesting that social disorganization resulting from mass incarceration may perpetuate firearm homicides. Targeted policies addressing mass incarceration and the disparities therein may be a means of reducing urban firearm homicides.

Introduction

Firearm-related injury remains a substantial burden in the US with over 45 000 people dying from these injuries in 2020.¹ While rates of firearm homicide had been decreasing between 1990 and 2010, and suicides historically have represented two-thirds of all gun deaths, there has been an increase in the last decade, and homicides now make up approximately 45% of all firearm-related deaths.² There are wide disparities in age and race of those affected.¹ Young Black men are significantly more likely to be exposed to firearm homicide,³ which are themselves hyperconcentrated in urban areas with causal factors based in structurally racist practices, including redlining and downstream concentrated disadvantage.^4,5,6

There has been much discussion, particularly within criminology and sociology research, as to the utility of imprisonment in preventing firearm assaults and homicides. With the rise of mass incarceration during the Reagan era and the War on Drugs, incarceration rates increased 5-fold between 1975 and 2009, with nearly 1 in 100 adults behind bars at the peak.⁷ There have been wide disparities among those incarcerated, largely due to targeted policing among racially and ethnically minoritized communities and differential sentencing.⁷ Previous studies have shown that Black men have incarceration rates far exceeding those of their White counterparts, with high school–educated Black men having a 1 in 3 chance of being incarcerated during their lifetime.⁷ This statistic increases to 2 in 3 Black men incarcerated among those that did not finish high school.⁸ By contrast only 15% of White men with less than a high school degree are incarcerated in their lifetime.⁹

Mass incarceration has substantial deleterious effects on the family unit and on economic well-being.¹⁰ Targeted policing and drug offenses impact already marginalized individuals and communities, stripping families of a source of income, which remaining family members are often unable to allay themselves.¹¹ Furthermore, felony convictions can prevent these men from obtaining a job after release from prison. Given the high number of arrests among Black men, there is unsurprisingly a large number of Black children that are affected. Among those Black children born in 1990, nearly 1 in 4 had a father go to prison.¹² Having incarcerated parents is itself associated with increased risk for behavioral issues,¹⁰ but can also leave an economic need within families which, due to limited job and education prospects, can often only be alleviated through extralegal measures.^13,14

Multiple theoretical models help to explain the association of incarceration with firearm violence. One is incapacitation and deterrence, whereby incarceration prevents future violence by taking offending criminals off the street.¹⁵ Another is social disorganization whereby mass incarceration targeting vulnerable communities leads to social disorder, socioeconomic instability, and ultimately increasing violence in those same communities.¹⁶

The current study sought to understand the association of race-specific incarceration with race-specific rates of shooting in Chicago, Illinois, a racially and economically segregated city. We further sought to understand the downstream effects on the family unit and socioeconomic vulnerability.

Methods

Data Sources

This cross-sectional study was deemed exempt from review by the Boston University institutional review board because it did not constitute human participants research and used deidentified data in accordance with the Common Rule. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Homicide data for the city of Chicago were obtained from The Trace, a news source dedicated to gun violence reporting.¹⁷ As part of their journalism, data were obtained through Freedom of Information Act requests to law enforcement agencies and subsequently released for research use. Data were obtained by our group in 2020, and data files provided a wide range of variables for different cities. For the purposes of this study, we used all available data for Chicago, which included all homicides in the city between January 1, 2001, and August 31, 2019, for all means of homicide. These data are reflective of all incidents that are reported by the Chicago police department and are subject to the biases of police reporting. Other variables included address of the incident, race, sex, and method of homicide (eg, shooting or stabbing). Incidents were limited only to homicides involving a firearm. We further limited incidents to individuals with listed race of Black, Hispanic, and White. Race in this data was collected through officer reports and may not be the same as self-reported race among the individuals.

Census tract level data were obtained from various sources. Population measures (Black, Hispanic, and White populations) were obtained at the census block level from the 2010 decennial census because yearly estimates for this geographic subdivision would be unstable.¹⁸ Social vulnerability was measured by the Social Vulnerability Index (SVI) from the Centers for Disease Control and Prevention.¹⁹ The SVI measure was created from 2018 census tract level data quantifying 16 variables via calculated percentile rank of social vulnerability across 4 domains: socioeconomic status, household composition and disability, minority status and English-speaking language proficiency, and housing and transportation. Rankings are based on 4 categorical variables: very low, low, moderate, and high social vulnerability. This composite measure was chosen instead of its separate economic components, given concern for collinearity with the other variables of interest. Incarceration rates were obtained by race and ethnicity at the census tract level from the Opportunity Atlas.²⁰ Researchers with this group followed individuals born between 1978 and 1983 and reported the percentage of people within a particular census tract that were incarcerated in 2018 based on Internal Revenue Service tax records and US census records. Chicago census block shapefiles were obtained from the data portal through the city of Chicago.²¹ This incarceration rate does not reflect reasons for incarceration, but based on national data for US incarceration, we would expect the majority to reflect drug sentences.²²

Statistical Analysis

Data analysis took place from January to June 2023 using Stata software version 16 (StataCorp). Statistical significance was defined as a 2-sided P < .05.

Data Cleaning and Analysis

Homicide data were limited to only those involving a gun. The addresses associated with each incident was geocoded to an x,y point location using ArcGIS version 10.8 (Esri).²³ These were then spatially joined with census block shapefiles of Chicago to assign each homicide a census block designation. We chose to perform the analysis at the census block level to allow granularity in assessment of shooting incidents. Shooting incident numbers were tabulated for each census block, creating a data set of all census blocks in Chicago and the number of shooting homicides in the included time frame. Census tract level data for population indicators, SVI, incarceration, and single-parent status were linked to each census block in the city. Shooting rates were then calculated for each census block based on census block population.

First, we sought to understand the associations of different census tract level variables and shooting rates by race. We created crude univariate regression models between the census tract variables and the outcome of interest using a Poisson regression. The association of race-specific incarceration rates with firearm incidents was assessed. Incarceration rates were evaluated as tertiles of incarceration rates within Chicago, with the highest tertile compared with the lowest tertile. Next, single-parent household and SVI were studied as mediators to assess for an association of the exposure (incarceration rate) with the outcome (firearm incidence). If associated with both the exposure and the outcome, they were considered a mediator of the association and included in subsequent models.

Structural Equation Modeling

We used structural equation modeling (SEM) to understand the mediating association of included variables with the outcome of interest.^24,25 For the purposes of this study, we sought to understand the interplay between incarceration, social vulnerability, and single-parent families through a theoretical framework of social disorganization.¹⁶ In this framework, we hypothesize that incarceration would lead to a higher number of single-parent households and increase social vulnerability, thus increasing firearm violence for communities that have the highest rates of incarceration (Figure). SEM models were created assessing the association of race-specific incarceration rates with the corresponding race-specific shooting rates through the mediators of SVI and single-parent households if baseline associations were found. We calculated the direct effects between the exposure (X) and outcome (Y), exposure (X) and mediator (M), and between the mediator (M) and outcome (Y). We then calculated the indirect effects of the association between the exposure and the outcome through each mediator and the total effect, which is the direct effect between the exposure and the outcome added to the indirect effect of each mediator. Included variables were chosen in different time frames (ie, temporally distant from one another) in order to create a crude time-dependent model around the included exposure and mediating variables.

Figure. Directed Acyclic Graph of Exposures, Mediators, and Outcomes Evaluated in Structural Equation Models.

Exposures included race-specific incarceration rates and the outcome included race-specific firearm homicide rates. Mediators of the association of incarceration with firearm homicide included single-parent households and Social Vulnerability Index.

Results

In total, 46 312 census blocks were evaluated. The vast majority of census blocks had 0 shootings. The median rates of shootings for Black, Hispanic, and White incidents were 0 for all levels of SVI but were significantly different across different levels of SVI for Black and Hispanic shootings (Table 1). They were not significantly different for White shootings. Levels of incarceration were highest in the highest level of SVI for Black individuals (median [IQR] incarceration rate, 4% [2%-5%]; P < .001), Hispanic individuals (median [IQR] incarceration rate, 1% [0%-1%]; P < .001), and White individuals (median [IQR] incarceration rate, 1% [0%-2%]; P < .001). There were significant differences in the proportion of single-parent households with the highest level of SVI having the highest rate (median [IQR] percentage, 28.4% [19.7%- 37.9%]).

Table 1. Baseline Census Block Characteristics by Levels of the Social Vulnerability Index.

Characteristic	Social Vulnerability Index score					P value
	Overall	Very low	Low	Moderate	High
Shooting rate, median (IQR) [range]
Black	0.0 (0.0-0.0) [0.0-11.4]	0.0 (0.0-0.0) [0.0-4.6]	0.0 (0.0-0.0) [0.0-11.4]	0.0 (0.0-0.0) [0.0-4.3]	0.0 (0.0-0.0) [0.0-9.6]	<.001
Hispanic	0.0 (0.0-0.0) [0.0-19.2]	0.0 (0.0-0.0) [0.0-1.6]	0.0 (0.0-0.0) [0.0-5.4]	0.0 (0.0-0.0) [0.0-6.6]	0.0 (0.0-0.0) [0.0-19.2]	<.001
White	0.0 (0.0-0.0) [0-62.5]	0.0 (0-0) [0.0-0.1]	0.0 (0.0-0.0) [0.0-0.6]	0.0 (0.0-0.0) [0.0-11.4]	0.0 (0.0-0.0) [0.0-62.5]	.11
Incarceration rate, median (IQR) [range], %
Black	3 (2-5) [0-42]	2 (1-6) [0-22]	3 (2-7) [0-29]	3 (2-4) [0-42]	4 (2-5) [0-18]	<.001
Hispanic	1 (0-2) [0-10]	1 (0-2) [0-9]	1 (0-2) [0-6]	1 (0-1) [0-8]	1 (0-1) [0-10]	<.001
White	0 (0-1) [0-13]	0 (0-1) [0-10]	0 (0-1) [0-13]	0 (0-1) [0-12]	1 (0-2) [0-9]	<.001
Single parent, median (IQR) [range], %	18.1 (10.2-28.8)	6.73 (3.2-11.1)	10.2 (6.7-16.5)	17.2 (12.6-24.0)	28.4 (19.7-37.9)	<.001

When comparing exposures, outcomes, and mediators (Table 2), there was a significant association of high (vs very low) SVI with Black shooting rate (incidence rate ratio [IRR], 2.34; 95% CI, 1.59-3.42), Hispanic shooting rate (IRR, 3.27; 95% CI, 2.07-5.14), and White shooting rate (IRR, 49.0; 95% CI, 20.2-118.7). Similarly, proportion of single-parent households was significantly associated with Black shooting rate (IRR, 1.03; 95% CI, 1.02-1.04), Hispanic shooting rate (IRR, 1.05; 95% CI, 1.03-1.06), and White shooting rate (IRR, 1.09; 95% CI, 1.05-1.12). Census blocks in the highest tertile of Black incarceration were associated with a 70% higher rate of Black shootings (IRR, 1.70; 95% CI, 1.50-1.94). There was no association of Hispanic incarceration with Hispanic shooting rate (IRR, 0.98; 95% CI, 0.75-1.28), nor was White incarceration associated with White shooting rate (IRR, 1.13; 95% CI, 0.65-1.97). Given the lack of an association of incarceration with firearm incidents among Hispanic and White individuals, SEM models were not created.

Table 2. Crude Associations of Census Tract Variables With Race-Specific Shooting Rates in Univariate Regression Models.

Characteristic	Shooting rate by race, IRR (95% CI)a
	Black	Hispanic	White
Social Vulnerability Index score
Very low	1 [Reference]	1 [Reference]	1 [Reference]
Low	1.05 (0.56-1.97)	1.02 (0.48-2.14)	1.36 (0.61-3.03)
Moderate	1.01 (0.67-1.52)	1.52 (0.89-2.58)	22.7 (10.2-50.7)b
High	2.34 (1.59-3.42)b	3.27 (2.07-5.14)b	49.0 (20.2-118.7)b
Race-specific incarceration rate, tertile
1	1 [Reference]	1 [Reference]	1 [Reference]
2	1.04 (0.91-1.18)	1.12 (0.91-1.38)	0.68 (0.39-1.16)
3	1.70 (1.50-1.94)b	0.98 (0.75-1.28)	1.13 (0.65-1.97)
Single parent	1.03 (1.02-1.04)b	1.05 (1.03-1.06)b	1.09 (1.05-1.12)b

Abbreviation: IRR, incidence rate ratio.

^a Rates per 10 000 people.

^b P < .001.

SEM: Black Shooting Rate

In assessing the association of incarceration with shooting rates, the incident rate of Black shootings was 67% higher (IRR 1.67; 95% CI, 1.54-1.80) in census blocks with the highest tertile of Black incarceration compared with those in the lowest tertile. When accounting for mediators of the association, the direct effect of shootings in the highest tertile of Black incarceration was 45% higher compared to the lowest (IRR 1.45; 95% CI, 1.42-1.49) (Table 3). Both SVI and single-parent households were assessed as mediators of the association. The indirect effect of single-parent households on the association of incarceration with shootings resulted in a 14% increase in shooting rates (IRR, 1.14; 95% CI, 1.12-1.15) and accounted for 23% of the mediation of the association of race-specific incarceration with firearm incidents. There was no indirect effect of SVI in the association of race-specific incarceration with firearm incidents (IRR, 1.01; 95% CI, 0.99-1.03). We further showed that the combined mediation of single parent households and SVI between incarceration and incident rate of shooting was significant and was associated with a 3% increased incident rate of shooting (IRR, 1.03; 95% CI, 1.01-1.05), which accounted for 5% of the total mediation of the association.

Table 3. Structural Equation Model Assessing the Association of Black Incarceration Rates (Measured as Highest vs Lowest Tertile of Incarceration) With Census Block Rates of Shootings of Black Individuals.

Interaction	Black shooting rate per 10 000 people (blocks with the highest level of incarceration vs those with the lowest), IRR (95%CI)			Mediation, %
	Direct effect	Total effect	Indirect effecta
High incarceration (vs low incarceration)	1.45 (1.42, 1.49)b,d	1.72 (1.65, 1.79)c,d	NA	NA
SVI (high vs very low)
Exposure × mediator	Correlation coefficient (95% CI), 0.22 (0.10-0.34)d	NA	1.01 (0.99-1.03)	NA
Mediator × outcome	Correlation coefficient (95% CI), 1.06 (1.01-1.22)e	NA
Single-parent household (continuous)
Exposure × mediator	Correlation coefficient (95% CI), 4.11 (3.75-4.48)d	NA	1.14 (1.12-1.15)d	23
Mediator × outcome	Correlation coefficient (95% CI), 1.03 (1.03-1.03)e	NA
Incarceration × single parent × SVI	NA	NA	1.03 (1.01-1.05)e	5

Abbreviations: IRR, incidence rate ratio; NA, not applicable; SVI, Social Vulnerability Index.

^a Indirect effect between exposure, mediator, and outcome. Indirect effects of each mediator and percent mediation are calculated for percentage of single-parent households and social vulnerability.

^b Direct effects between the exposure and outcome. Direct effects assess the association between incarceration and firearm homicide rates, accounting for indirect effects.

^c Total effect between the exposure and outcome. Total effect is the combination of direct and indirect effects in the causal pathway.

^d P < .001.

^e P < .05.

Discussion

Taken together, this cross-sectional study highlights the association of incarceration and social instability with rates of firearm violence in Chicago. Our findings demonstrated that Black incarceration rates were positively associated with Black firearm homicide rates, but the same was not true for Hispanic or White communities. We went on to show that this association was mediated through social vulnerability and single-parent households within these same geographic areas, which seems to reinforce theories of social disorganization as an important root of firearm violence and homicide in Chicago.¹⁶ Through a critical race lens, these findings highlight the racialized impact of incarceration and further explore the downstream socioeconomic and family effects that perpetuate violence in these communities.

Rates of incarceration have climbed over the last half century, with a 5-fold increase in incarceration rates between 1975 and 2009. The result is that among individuals in the US born in the 1990s, 29% of Black individuals can expect to serve a prison sentence compared with 16% of Hispanic individuals and 5% of White individuals.^7,12,26 This racially disparate mass incarceration preferentially targets Black men, particularly those with less than a high school degree.⁹ Our findings reinforce these statistics by showing higher rates of Black incarceration in the city of Chicago when compared with Hispanic and White rates of incarceration. Among Black individuals in Chicago, the highest levels of incarceration were associated with a 70% increased incident rate of shooting in the studied time period. For all shooting rates, social vulnerability and single-parent households were shown to be directly associated with higher rates of firearm violence, suggesting that socioeconomic factors contribute to higher rates of firearm homicides for all groups.

While previous studies¹⁵ have pointed to incapacitation and deterrence models as means of decreasing firearm violence by removing violent offenders from society, regardless of the subsequent impact this may have on the family unit or community as a whole, our findings lay doubt to the explanatory nature of this model. Previous studies⁶ looking at firearm homicides in metropolitan statistical areas have shown there to be convergence in the disparity between Black and White shooting rates with a concurrent widening of racial disparities in incarceration. This was used as evidence to reinforce deterrence models of incarceration and strengthen the carceral state; however, given the wide geographic measures and crude associations, the authors⁶ conceded that further geographically nuanced studies should be performed. Other studies have shown that lower levels of incarceration can reduce crime, but higher levels have less elastic associations with crime reduction and may actually potentiate crime.²⁷ Our SEM models incorporate shooting rates for Black, Hispanic, and White individuals, and study the neighborhood (census tract) level effects that lead to disparate rates of shootings. If incarceration was associated with decreased violence in communities under an incapacitation model, we would expect a decrease in firearm homicides in areas with higher rates of incarceration for all races. While limited to 1 city, we showed that higher rates of incarceration were associated with increased rates of firearm homicides for Black individuals, while there was no observed association for Hispanic or White individuals.

We went on to show that the association of incarceration with firearm homicide rates among Black individuals was mediated through the percentage of single-parent households, which collectively mediated approximately 30% of the interaction. SVI was found to have a lesser effect on this association and was only found to be a downstream mediator from single-parent households, suggesting that single-parent households and not incarceration alone may worsen socioeconomic disparities at the neighborhood level. These findings reinforce models of social disorganization that propose that by removing individuals from families and society, overwhelmingly Black men in the US, incarceration can disrupt ties within families and communities.¹⁶ When released from prison, already limited job opportunities within socially disadvantaged areas can become even more scarce due to felony convictions.^8,12 Studies have shown that having an incarcerated family member can reduce a household’s assets by 64% and increase debt by 85%.⁸ Similarly, single-parent households have been associated with behavioral issues for children and higher rates of youth homicide.²⁸ Interestingly, while social vulnerability was directly associated with firearm violence in isolation, it was not a direct mediator of the association of incarceration rates with firearm violence, which may reflect that incarceration is targeted within already socially vulnerable communities and does not contribute to worsening of this measure. It is important to note that while arrests and subsequent incarceration may occur for a wide variety of violent and nonviolent offenses, our incarceration data were unable to separate these types of offenses at such a granular level. Regardless, under a social disorganization model, incarceration regardless of offense is the studied driving factor perpetuating crime. Taken together the socioeconomic impact on family units in already socially vulnerable communities that themselves have limited job prospects can lead to unequal means of support and means of income.²⁹ It is important to note that the direct effect of incarceration remained significant even after accounting for mediation, suggesting that there are other inherent factors to incarceration or other unmeasured mediators that are important in this interaction.

We did not find associations of incarceration with firearm homicides when looking at rates of incarceration among Hispanic and White individuals. However, we did show associations of social vulnerability and single-parent households in isolation with firearm homicides, suggesting that socioeconomic factors too mediate rates of firearm violence in these communities, but through a different avenue than in Black communities. We studied a wide year range in order to achieve adequate power to study the rare instances of White firearm homicides in Chicago. Despite the large time frame included, there was a significant association of SVI with White firearm homicides, but there was a wide confidence interval, suggesting these estimates may not be stable and further highlight the wide disparities in firearm homicide rates in Chicago and other urban areas across the country. Given the scope of the study, we did not further elucidate other avenues through which socioeconomic factors influence Hispanic and White firearm homicides, but further studies should be performed to elucidate these pathways. Incarceration statistics suggest that Hispanic communities were less affected by mass incarceration trends of the 1980s and 1990s, possibly because of the types of drug offenses that drove the levels of incarceration.³⁰ There may also be some protective effect of immigration and emigration within these communities that may provide some stability in the face of social disorganization secondary to incarceration.³⁰ While we did not find an association with incarceration, there are still disparities in firearm violence between Hispanic and White communities that may not be explained by incarceration, but that may still lie in socioeconomic and neighborhood-level factors.

Limitations

Our study has some important limitations. While using a causal method of understanding mediation, our data were retrospective and subject to unmeasured biases. Similarly, our data did not necessarily follow causal trends and were limited to available data. The incarceration rates come from the Opportunity Atlas, which is a snapshot in time of incarceration rates among a large cohort and may not represent the full extent of incarceration over the preceding decades; however, this is the most geographically nuanced measure to date. Similarly, the incarceration data do not provide offenses for which individuals are incarcerated, which limited nuanced understanding of the impact of violent vs nonviolent offenses. Our census variables, including SVI and single-parent households, represented a single year and we were unable to assess changes in these measures given the static nature of the analysis. Similarly, the studied time period was large so as to have enough power to study White shootings but could not accurately depict time-based changes in shootings.

Conclusions

Overall, our study illustrates the detrimental impact of mass incarceration on Black communities with a race-specific impact of incarceration on Black communities through a social disorganization perspective. Our results suggest that policies targeting reparative action around mass incarceration and dismantling of the carceral state can act as a platform with which to decrease firearm homicides in racially marginalized communities.

Supplement.

Data Sharing Statement