How do county-level ethno-racial composition and residential segregation influence pretrial decisions? Results from a multi-level,multi-decade study

Abstract

The American criminal legal system is characterized by deeply entrenched racial and ethnic disparities. Given that aggregate criminal legal system outcomes are a function of decisions made by police, prosecutors, judges, and other local actors, it is reasonable to expect that localized punishment decisions are contingent on the racial and ethnic dynamics within a community. However, there is a dearth of research examining the impact of community characteristics on pretrial case-processing outcomes. The current study explores the impact of county-level minority composition and segregation on pretrial decision-making. Drawing from a sample of 31,974 felony defendants across 64 large urban counties, we examine the relationships between community racial and ethnic composition and segregation on release on recognizance, bail denial, and bail amount decisions. Our findings reveal that minority composition and segregation directly influence court case processing, and they operate in contrasting ways across different decision points. The findings underscore the importance of community characteristics above and beyond individual-level factors in determining pretrial decisions.

Keywords

pretrial detection segregation judicial decision-making community context

Introduction

Racial residential segregation remains a defining characteristic of American cities (Massey and Denton, 1993; Rothstein, 2017). Research consistently finds that individuals residing within racially isolated communities of color face limited job opportunities (Leonard, 1987), unstable living conditions and resulting residential mobility (Charles, 2003), poorer health conditions (Williams and Collins, 2001), heightened probability of criminal victimization (Peterson and Krivo, 1993; Shihadeh and Flynn, 1996), and lackluster educational systems (Massey et al., 1987). Segregation, coupled with the racial and ethnic composition of communities, is frequently implicated as drivers of crime, criminal legal system involvement, and disparate criminal legal outcomes in the United States (Krivo et al., 2009).

The group threat perspective posits that residents of communities with large proportions of minorities will exhibit heightened fear of crime (e.g. Chiricos et al., 2001) and hold more punitive attitudes (e.g. Blumer, 1958; Bobo and Hutchings, 1996; Taylor, 1998) than those with smaller minority proportions. These attitudes impact the decisions of criminal legal actors (Pickett, 2019), who are sensitive to public opinion amidst concerns over reelection (e.g. Huber and Gordon, 2004) and institutional legitimacy (e.g. Casillas et al., 2011). Such communities are frequently characterized by a greater reliance on formal social control (e.g. Blalock, 1967; see Vogel and Messner, 2024 for a recent review), which perpetuates the “racialized social system” (Krivo et al., 2009: 1765). As Krivo and colleagues (2009: 1767) argue, “practices are organized to produce and reinforce racial order…through racial inequality in the application of justice…to reproduce White privilege and reinforce the status quo of minorit[ies].” Scholars consistently find positive associations between minority population size and criminal legal system capacity-building, whether it be through changes to police force sizes (Stults and Baumer, 2007) or correctional budgets and incarceration (Greenberg and West, 2001; Williams and Campbell, 2021). More recently, researchers have begun examining relationships between aggregate racial and ethnic composition and individual-level punishment decisions, most notably sentencing (e.g. Beckman and Wang, 2022; Donnelly, 2022; Wang and Mears, 2010; Zvonkovich and Ulmer, 2023).

While punishment decisions likely vary based on the racial and ethnic contexts of local communities, extant research on race and punishment is limited in important respects. First, few studies of individual-level case-processing decisions distinguish between minority population composition and segregation. This is problematic, as these community features may operate differently depending on how they are perceived to threaten social boundaries. For example, while the proportions of minorities in a county can increase, the perceived threat may be minimal if minorities are concentrated in highly segregated neighborhoods and have limited social interaction with majority group members.

Further, while some studies have explored the dynamic relationship between community context and early stages of criminal case prosecution (see Donnelly and Asiedu, 2020; Pinchevsky and Steiner, 2016; Sutton, 2013; Williams, 2016; Williams and Rosenfeld, 2016; Wooldredge, 2007), few have examined the impact of segregation on pretrial detention. This oversight is important as research shows significant (and negative) impacts of pretrial detention on downstream decision-making (for a recent meta-analysis, see St Louis, 2024). For instance, defendants held prior to trial face an increased likelihood of incarceration and harsher sentencing (Kurlychek and Johnson, 2019; St Louis, 2024). Given the wealth of research showing that criminal legal actors are sensitive to features of their localities, such as population composition, economic conditions, sociopolitical culture, and crime (Baumer and Martin, 2013; Beckman and Wang, 2022; Britt, 2000; Eitle et al., 2002; Lynch, 2011; Sutton, 2013) and that pretrial decision-making varies widely across jurisdictions (Baumer, 2013; Britt, 2000), we should reasonably expect minority composition and segregation to account for at least some of the county-level variation in pretrial judicial decision-making, which may ultimately lead to more punitive sentencing outcomes. Drawing on individual-level court processing data from several dozen large urban US counties, coupled with county-level indicators of ethno-racial composition and segregation, we draw attention to the nuances between differing measures of ethno-racial threat and their effects on pretrial detention decisions.

Theoretical perspective

Racial composition may contribute to county-level variation in pretrial outcomes by operating on mechanisms of perceived threat (Blalock, 1967; Liska, 1992). Group threat theory anticipates that large or growing minority populations threaten the perceived status quo, leading to increased prejudice and attempts by the dominant and powerful majority group to control minorities (e.g. Blalock, 1967; see Vogel and Messner, 2024 for review). In this framework, the term “majority” typically refers to White Americans with disproportionate levels of privilege and power, while “minority” comprises members of historically marginalized racial and ethnic groups. High unemployment and inequality may exacerbate punitive practices, as growing populations of idle citizens with ample free time and scarce resources may lead to increased crime (Chambliss and Seidman, 1971; Liska, 1992). Similarly, when groups are isolated from each other (i.e. segregated), even within the same community, they pose little threat to the shared interests of majority group members (Spitzer, 1975). It is only when the minority community becomes increasingly visible that status consciousness should emerge, thereby increasing the perceived threat. From the same vantage point, we might expect the threat to be relatively low when communities are well-integrated, as racial prejudice and animus are likely mitigated by familiarity and frequent contact (Allport, 1954; Pettigrew and Tropp, 2006). Others, such as Wacquant (2001), have argued the opposite—that segregation should worsen punishment because it cultivates insecurity and reinforces perceptions of symbolic threat from minority groups.

Empirical research has found that growing minority populations can increase racial prejudice and punitive sentiment toward minorities (Bobo and Hutchings, 1996; Taylor, 1998; Welch et al., 2011) and that growth in non-White populations is strongly correlated with fear of crime (Chiricos et al., 2001). In their recent review of group threat and social control, Vogel and Messner (2024) point to important and nuanced findings regarding minority group size increases and fear among dominant group members. For instance, research by Piatkowska and Messner (2022) found some support for the power-differential hypothesis, which states that increasing minority group size—and the power increases that result from it—can actually cause dominant group members to fear expressing hate toward minorities. Research also suggests that minority group size affects decision-making by local criminal legal actors. Police, judges, and prosecutors respond punitively to growing minority populations; and in places where minority populations are large or growing, criminal legal systems tend to expand. For instance, higher proportions of Black residents are associated with police force size increases (though this relationship is curvilinear—see Greenberg et al., 1985; Stults and Baumer, 2007), elevated arrest rates (Liska, 1992; Liska and Chamlin, 1984; Parker et al., 2005) and increased criminal legal system operating budgets (Williams and Campbell, 2021).

Since aggregate criminal legal outcomes reflect decisions by police, prosecutors, judges, and other local actors (Forman Jr., 2017; Lynch, 2011), it follows that localized decisions, such as sentencing, should vary based on the racial and ethnic contexts of the communities in which they are made. The limited research that has examined the influence of minority population size on individual sentencing outcomes has produced equivocal findings (Britt, 2000; Fearn, 2005; Helms and Jacobs, 2002; Ulmer and Johnson, 2004; Wang and Mears, 2015). While some find no significant effects of county-level racial composition on individual-level sentencing (e.g. Fearn, 2005; Helms and Jacobs, 2002), Britt (2000) shows that individuals charged in counties with large Black populations face increased odds of incarceration. This suggests that where perceived threat should be high, punitiveness tends to increase, and members of all racial groups face harsher penalties.

It is important to note that while research has focused primarily on formal responses to changes in the relative size of Black populations, responses may differ for Hispanic populations (Eitle and Taylor, 2008; Markert, 2010; Zvonkovich and Ulmer, 2023). For instance, Ouellette and Applegate (2023) find that county-level jail admission rates vary by county-level racial and ethnic population sizes. Percentage Black and jail admission rates resemble a U-shaped curvilinear pattern in which jail admissions decrease until percentage Black reaches a tipping point and admissions begin increasing, while there is no significant effect for percentage Hispanic. However, the non-linear relationship between jail populations and percent Black and percentage Hispanic operates in contrasting ways. Jail populations were little affected by smaller percentages of minority populations, but once a tipping point was met, percent Black models an accelerating pattern, with jail populations increasing significantly. In contrast, percentage Hispanic modeling a decelerating trend, with jail populations decreasing significantly (see also Wang and Mears 2010). This suggests that punishment is highest in places with very large and very small Black populations. Conversely, the trend for Hispanic populations suggests that punitiveness increases until the Hispanic population reaches a critical mass, after which the reliance on formal control begins to level off and recede.

The mixed findings related to racial threat and punishment decisions might be attributed to issues of how ‘threat’ is measured (Feldmeyer and Cochran, 2018; King and Light, 2019; Ulmer, 2012). For one, researchers often interpret “the correlations [between racial composition and punishment] as consistent with group threat theory,” despite their reliance on different measures of racial composition and vastly different conclusions regarding their effects (King and Light, 2019: 405). Researchers also tend to rely on standard linear measures of ethno-racial composition without attending to “non-linear effects or tipping points” (Ulmer, 2012: 30). Relatedly, group threat research is distinguished from research explaining racial tolerance, such as contact theory (Feldmeyer and Cochran, 2018). This prevents us from fully understanding the effects of measures that capture potential interactions, such as residential segregation and integration, and their potential non-linear relationships with punishment (Feldmeyer and Cochran, 2018).

These insights provide some theoretical guidance to anticipate the nature and direction of racial and ethnic residential segregation on pretrial decisions. On the one hand, the group threat perspective might assume that punishment should be relatively low when segregation is high. High levels of segregation translate to limited interaction between members of different racial and ethnic groups. When contact is minimized, the perceived threat of one group encroaching on another’ symbolic, spatial, or cultural boundaries is minimal. By the same token, Allport's (1954) contact theory suggests that frequent contact should increase tolerance, meaning that perceived threat, and thereby punitiveness, should also be relatively low in counties characterized by a high degree of racial and ethnic integration. It is less clear how pretrial decisions will vary between communities characterized by very high and very low levels of segregation. The neighborhood effects literature provides some direction here. Green and colleagues pioneering study of NYC neighborhoods found that as minority presence increased within predominantly White neighborhoods, so too did crimes directed at minorities (1997; see also Lyons, 2007). Legewie and Schaffer (2016) and Vogel et al. (2023) similarly report that rates of interpersonal violence are highest in the transitional areas between racially ethnic and homogenous communities. These literatures underscore a curvilinear relationship between residential integration and punishment, with social control being lowest in poorly integrated and highly integrated communities and higher elsewhere.

Background literature

While researchers have recently begun to theorize and empirically investigate how the social context surrounding criminal cases impacts case processing outcomes, including pretrial detention and bail, they traditionally emphasized the role of case-level legal and extralegal factors. Defendants who face serious charges and those with lengthy criminal records receive more punitive sanctions during pretrial stages (Demuth, 2003; Wooldredge, 2012), regardless of their probation or parole status and whether they have previously failed to appear at trial (Demuth, 2003; Schlesinger, 2005). Research also points to the importance of extralegal factors, with studies generally finding that being young and male leads to more pretrial restrictions (Wooldredge, 2012). While race and ethnicity affect various criminal legal system outcomes, findings regarding their effects on pretrial outcomes have been mixed, particularly when legal factors, such as offense severity and a defendant's prior criminal record, are considered (Freiburger and Hilinski, 2010; Freiburger et al., 2010; Stolzenberg et al., 2013). For instance, research by Wooldredge (2012) finds that while African Americans receive higher bond amounts, much of this relationship can be explained by defendants’ current charges and prior records.

Scholars have further explored how features of the broader social context amplify the effects of legal and extralegal factors on case processing decisions, including pretrial detention, bail, and sentencing (Ulmer, 2012; see also Baumer and Martin, 2013; Beckman and Wang, 2022; Britt, 2000; Eitle et al., 2002; Hood and Schnieder, 2019; Johnson, 2006; Sutton, 2013). For example, Baumer and Martin (2013) found that punishments for murder depend on an area's levels of social capital, religious fundamentalist values, support for punitive sanctions, and fear. Economic factors (Pinchevsky and Steiner, 2016; Sutton, 2013; Williams and Rosenfeld, 2016) and levels of household disadvantage (Pinchevsky and Steiner, 2016) have been associated with worse pretrial detention outcomes, such as higher bail amounts. Racial and ethnic minorities adjudicated in counties with high concentrations of poverty receive particularly negative outcomes, including a lower likelihood of receiving bail (Sutton, 2013).

Research on ethno-racial composition and segregation has been limited, with most studies limiting their focus to linear relationships between racial composition measures and downstream case processing outcomes. Research on the effects of community racial composition on sentencing is mixed (King and Light, 2019; Ulmer, 2012). Some research suggests that areas with larger minority populations are more likely to incarcerate (e.g. Weidner et al., 2005) or mete out harsher sentences, particularly to minority defendants (Ulmer and Johnson, 2004). Others find no relationship between racial composition and incarceration or sentencing (e.g. Weidner and Frase, 2003) or evidence that counters racial threat (e.g. Britt, 2000). As noted above, the equivocal nature of these research findings highlights the need to move beyond simple measures of racial composition in favor of more nuanced measures of racial threat, both theoretically and empirically (King and Light, 2019; Ulmer, 2012). As an illustrative example, Beckman and Wang (2022) found that the relationship between sentencing and segregation was nonlinear—sentencing decisions tend to be most lenient in highly segregated communities—suggesting that indicators of racial threat may operate in counterintuitive ways to influence case outcomes.

We have a limited understanding of the relationship between racial threat and upstream criminal legal decisions, which is problematic given the findings linking pretrial detention to more punitive sentencing. To date, only one study has explored the role of county-level racial composition on pretrial detention. Hood and Schneider's (2019) work finds county variation in reliance on pretrial detention; and that contextual factors, rather than legal or extralegal factors, are associated with bail changes over time. The researchers find political, economic, and racial composition influences on pretrial outcomes; with both pretrial release probability and bail amounts being lowest in places with the largest Black populations. Overall, the relationships between threat measures and case processing outcomes appear to be complex and outcome-specific, underscoring the need for further research in understudied areas such as pretrial detention and bail (Hood and Schneider, 2019).

Current study

Research on racial composition and case processing has primarily focused on the roles that judges play in sentencing defendants, with limited attention given to pretrial outcomes. While the ultimate power to sentence resides with the judge, an overwhelming majority (>95%) of sentencing decisions are made through plea bargains offered by prosecutors. (Davis, 2007; Reaves, 2013). Sentencing decisions by judges may also be circumscribed by state-level sentencing guidelines (Nicholson-Crotty, 2004), whereas prosecutors can avert such constraints through downward departures in charges (Ulmer et al., 2007). As Kramer and Ulmer (2002) contend, the tools provided to a prosecutor to grant a downward departure can be viewed as a local court's ability to “correct” sentences (i.e. punishments) that do not conform to their primary concerns. While discretionary issues such as these may attenuate the roles judges play in sentencing, judges wield considerably more discretion at the pretrial phase (Dhami, 2005).¹ Assuming judges are receptive to minority group size and make decisions consistent with group threat theory, these patterns are more likely to emerge during the case-processing stages characterized by the most judicial discretion. County-level racial and ethnic composition can be expected, then, to impact pre-sentencing decisions more directly. In line with the above literature, we hypothesize that counties will have more punitive pretrial decision-making when Black populations are relatively small and relatively large (Hypothesis 1: U-Shaped Association). Conversely, we expect punitive decision-making to increase until the Hispanic population has reached a tipping point, after which point the association will reverse (Hypothesis 2: Inverted U-Shaped Association).

Group threat theory predicts that punitive attitudes will increase in response to growing contact between minority and majority populations (e.g. Duxbury, 2021). Similarly, group threat perspectives assume that segregation minimizes perceived threat, which should in turn suppress social control (Spitzer, 1975). As such, we hypothesize that residential segregation will be inversely related to punitive pretrial decision-making to a point. Drawing from the contact hypothesis, we expect that punitiveness will begin to decrease once levels of segregation, and thereby contact between Whites and minorities decrease to a critical limit, suggesting an inverted U-shaped relationship between segregation and punitive pretrial decisions. We have no reason to suspect this nonlinearity to operate differently for the Black and Hispanic populations.

Data and methods

We draw on data from the State Court Processing Statistics (SCPS), compiled by the US Bureau of Justice Statistics. These data include case-processing information at the individual level for felony defendants processed through a range of large urban counties for the even years from 1990 through 2006 and 2009.² These data were collected utilizing a two-stage stratified sampling procedure: the first stage selected 40 of the most populous 75 counties, and the second stage was used to systematically select a sample of felony case filings for the county.

The original dataset included 151,461 felony cases. Given the scope of the current study, we limited our sample to felony case filing data in years that correspond with the decennial census (1990, 2000, and 2009), resulting in 45,066 felony case filings nested within 67 counties. After excluding felony case filings in which key variables were missing (N = 6311 [14%]) and those in which the defendant was under the age of 16 (N = 84 [<0.2%]), missing age or gender data (N = 274 [<1%]), or classified as a race or ethnicity other than non-Hispanic Black, non-Hispanic White, or Hispanic (N = 6507 [14%]); the final analytic sample consisted of 31,974 individual case filings (71% of the total sample across the three time points). Finally, to explore the influence of county-level factors on pretrial detention outcomes, we culled data from the U.S. Census Bureau, Bureau of Labor Statistics, U.S. Department of Justice, and Federal Bureau of Investigations. We removed from the sample three counties missing data on important covariates, resulting in a total sample of 64 counties.

Dependent variables: Release on own recognizance, bail denial, and bail amount

Pretrial detention is a trifurcated judicial decision-making process (Goldkamp and Gottfredson, 1979). Judges are first tasked with deciding whether they will release a defendant back to the community on their own recognizance (ROR). ROR is measured in the current study as a dichotomous variable (1 = yes; 0 = no).

If a defendant is not released on their own recognizance, a judge must decide whether to deny bail to preventatively detain them. Under certain conditions, preventative detention is deemed necessary for preventing someone from failing to appear in future court hearings and/or engaging in criminal activity before trial (Klein and Wittes, 2011). Denied bail, a dichotomous variable (1 = yes; 0 = no), is used to capture the decision to outright deny any type of release for a defendant, thereby reflecting detention until trial.

Finally, for cases in which ROR is denied but bail is not, judges must decide an appropriate bail amount. Bail amount is meant to be an objective measure of the amount of monetary collateral that will incentivize a defendant to attend trial (if released on bail pretrial). It indicates a judge's perceptions of a defendant's dangerousness, as higher bail amounts are typically tied to those with more serious charges (Wooldredge et al., 2015) and longer criminal records (Stolzenberg et al., 2013). We rely on a continuous measure of bail amount that has been inflation-adjusted to 1990 dollars. Observations for bail amount are conditional on not receiving a ROR or having bail denied; that is, only cases in which a bail amount is offered are included in this measure. This variable is log-transformed to correct for positive skew (Fox, 2015).

Individual-level independent variables

We control for a variety of defendant and case-related factors. Age is measured as a continuous variable capturing the defendant's age at time of arrest. Gender is measured as a dichotomous variable (1 = male; 0 = female). Race and Ethnicity are measured as three dichotomous indicators: non-Hispanic Black, non-Hispanic White (reference category), and Hispanic only (hereafter Black, White, and Hispanic). We include indicators of a defendant's current charge (crime type), as well as their number of current charges. Crime type includes a set of dichotomous variables that account for the most serious charge a defendant is facing (violent [reference category], property, drug, or public order). Number of current charges is a continuous variable indicating the number of charges a defendant is facing for the current arrest, ranging from one to eight or more.

The final set of individual-level variables includes factors related to the prior record of the defendant: prior felony arrests and convictions, prior failures to appear (FTA), and active criminal legal system status. Prior felony arrest and prior felony convictions are continuous variables that range from zero to 10 or more. Prior failures to appear is a dichotomous variable indicating whether a defendant has failed to appear before the court for a prior arrest or charge (1 = yes; 0 = no). Active criminal legal system status is a dichotomous indicator of whether a defendant had a current active status at the time of their arrest, including whether the defendant was arrested for a prior offense while currently on pretrial release; was on probation, parole, or in a diversion program; was in custody; was a fugitive of the state; or had some other active status (1 = yes; 0 = no). Lastly, the model includes dichotomous variables for years 2000 (1 = yes; 0 = no) and 2009 (1 = yes; 0 = no), treating 1990 as the reference year, to account for secular changes that might affect judicial decision-making over time.

Community-contextual factors

Our primary focus is on the influence of county-level ethno-racial composition and segregation on bail-related decisions. Racial and ethnic demographic data are drawn from the US Census. Following prior research, we include county-level percentage Black and percentage Hispanic to measure the compositions of non-Hispanic Blacks and Hispanics in each county. We also include measures of Black-to-White and Hispanic-to-White segregation using the index of dissimilarity (D), defined as the proportion of one group (race/ethnicity) that would need to move into the other group's census tract to have even distribution throughout the larger geographic area. D ranges from zero to one, with one reflecting complete residential segregation. For ease of interpretation, we transform the measure into a percentage (D*100).³ In order to test our hypotheses regarding the nonlinear effects of racial composition and segregation on pretrial outcomes, we include quadratic terms in the regression. Therefore, we include squared terms for all four measures: percentage Black squared, percentage Hispanic squared, Black-to-White segregation squared, and Hispanic-to-White segregation squared.

We further control for county factors that have theoretical grounding in shaping punishment. Because crime concerns influence punishment decisions, we include a measure of a county's violent crime rate by combining murder, rape, robbery, and aggregated assault per 1000 residents (Steffensmeier et al., 1998), as provided by the Uniform Crime Reports. Higher levels of economic inequality and unemployment may signal resource scarcity, leading to greater levels of control (Chambliss and Seidman, 1971; Liska and Chamlin, 1984). To control for these factors, we include measures of economic inequality and unemployment. Data from the U.S. Census Bureau are used to calculate the GINI coefficient, which serves as a measure of economic inequality and points to income disparities across counties This measure is estimated using the user-written Stata command fastgini (Sajaia, 2007). Unemployment rate represents the potential threat of idle persons (Aaltonen et al., 2013), and is measured as the percentage of the workforce that is not currently employed using data from the Bureau of Labor Statistics.

Analytical strategy

We employ multilevel linear and logistic regression models to assess the influence of racial composition and segregation on pretrial decision outcomes. Such multi-level modeling strategies are useful in accounting for the nested structure of the data, in which cases are nested within counties.⁴ All models are estimated using the mixed and xtmelogit routines in Stata (V. 16).

The analyses unfold over several stages. The first step involves running unconditional models that highlight any significant differences between counties (available upon request). Assuming that county-level factors influence pretrial detention decisions, between-county variance should exist. The second stage therefore employs several random-intercept regressions. The first explores the influence of individual-level (i.e. level one) factors on ROR (Model 1), bail denied (Model 2), and logged bail amount decisions (Model 3), followed by models that assess the influence of county-level factors on pretrial detention decisions while controlling for individual-level factors. The first set of models explores the linear and non-linear effects of county-level racial composition and segregation on decisions of ROR (Model 4), bail denial (Model 6), and logged bail amount (Model 8), controlling for violent crime and economic indicators; followed by models assessing the linear and non-linear effects of county-level ethnic composition and segregation on ROR (Model 5), bail denial (Model 7), and logged bail amount (Model 9).⁵

Results

Table 1 provides descriptive statistics for the variables used in the study. The sample is predominately male (83%). Forty-five percent of defendants are non-Hispanic Black, 32% are non-Hispanic White, and 23% are Hispanic. Defendants have, on average, just under three prior felony arrests and just over one prior felony conviction. Approximately 32% have an active criminal legal system status, and 30% have previously failed to appear in court. The average number of current charges is just over two, with the highest frequency of most serious current charge being a felony drug charge (34%). Approximately 13% of defendants were released on their own recognizance, and 5% were denied bail. By race/ethnicity (results suppressed for parsimony, available by request), non-Hispanic Whites received ROR in just over 15% of cases, followed by Hispanic defendants at about 14%, and non-Hispanic Black defendants in 11% of cases. Further, non-Hispanic White defendants were denied bail in four and a half percent of cases, followed by non-Hispanic Black and Hispanic defendants being denied bail in five and a half and 6% of cases, respectively. When offered bail, defendants faced an average bond amount of approximately $28,500, with large (and statistically significant) disparities between defendants based on race/ethnicity. On average, non-Hispanic White defendants received bail amounts of $21,600, followed by non-Hispanic Black defendants at $26,200, and Hispanic defendants receiving the highest bails averaging just under $43,000.

Table 1.

Descriptive statistics for individual- and county-level predictors.

Individual-level (N = 31,974)	Mean	S.D.	Min	Max
Gender (1 = male, 0 = female)	0.83	—	0	1
Age	30.76	10.56	16	87
Non-Hispanic Black defendant (1 = yes; 0 = no)	0.45	—	0	1
Hispanic defendant (1 = yes; 0 = no)	0.23	—	0	1
Non-Hispanic White defendant (1 = yes; 0 = no)	0.32	—	0	1
Most serious current charge: violent (1 = yes; 0 = no)	0.24	—	0	1
Most serious current charge: property (1 = yes; 0 = no)	0.31	—	0	1
Most serious current charge: drugs (1 = yes; 0 = no)	0.34	—	0	1
Most serious current charge: public order (1 = yes; 0 = no)	0.11	—	0	1
Total current charges	2.19	1.63	1	8
Prior felony arrests	2.98	3.59	0	10
Active criminal legal system status (1 = yes; 0 = no)	0.32	—	0	1
Prior failure to appear (1 = yes; 0 = no)	0.30	—	0	1
Prior felony conviction	1.26	2.21	0	10
Year 1990 (1 = yes; 0 = no)	0.18	—	0	1
Year 2000 (1 = yes; 0 = no)	0.38	—	0	1
Year 2009 (1 = yes; 0 = no)	0.44	—	0	1
Outcome
Release on own recognizance (ROR (1 = yes; 0 = no))	0.13	—	0	1
Denied bail (1 = yes; 0 = no)	0.05	—	0	1
Ln bail amount^a	8.86	1.57	0	15.79
County factors (N = 64)
Violent crime rate	7.72	5.98	0.67	23.87
Unemployment rate	6.04	2.18	2.59	12.97
Gini coefficient	0.25	0.07	0.15	0.46
Index of Black-to-White dissimilarity	0.59	0.14	0.31	0.84
Percentage Black	15.44	13.07	1.20	66.07
Index of Hispanic-to-White dissimilarity	0.41	0.11	0.18	0.68
Percentage Hispanic	15.51	14.26	0.32	80.52

Note: Standard deviations for dichotomous variables are suppressed.

N = 21,173.

The average county in the sample has a violent crime rate of nearly eight per 1000 residents, faces moderate levels of economic inequality, and has an unemployment rate of about 6%. The composition of Black residents in the sample counties range from just over 1% to 66%, with the average county being about 15% Black. The mean Black-to-White index of dissimilarity was approximately .60, meaning that 60% of Black residents would need to relocate into predominately White census tracts to have complete residential integration. County-level percentages of Hispanics range from less than 1% to just over 80%, with an average of 15 and a half percent, and the mean Hispanic-to-White index of dissimilarity was .41.

The unconditional models highlight the between-county variation in pretrial detention and bail decisions (results available upon request), which can be used to compute intraclass correlation coefficients (ICCs). The results suggest that nearly 36% of the total variation in bail amount is due to differences in average bail amounts across counties (ICC = 0.363). Further, these models suggest that the use of ROR (ICC = 0.489) and denial of bail (ICC = 0.304) vary significantly across counties. However, it is worth noting that ICCs in multilevel logistic models are not very useful in determining between-group differences, as the level one variation is taken to be the standard logistic distribution ( $π^{2} / 3$ ). Thus, while the analysis allows for a calculation of the ICC, interpretation should be done with caution, as the ICC's “meaningfulness depends on the validity of the underlying distributional assumptions” (Johnson, 2006: 279).

Individual-level predictors

Table 2 presents the individual-level models predicting ROR, bail denial, and logged bail amount. The results demonstrate that male defendants are significantly less likely to be given ROR and more likely to be denied bail, and they receive bail amounts that are approximately 31% higher than females. Relative to White defendants, Black defendants are no more or less likely to be given ROR or denied bail; but Hispanic defendants are significantly less likely to be given ROR and more likely to be denied bail than Whites. When bail is offered, Black and Hispanic defendants receive bond amounts that are 10% and 13% higher than White defendants, respectively.

Table 2.

Hierarchical linear and logistic regression estimates predicting the individual-level effects on ROR, bail denied, and bail amount.

	Model 1 ROR		Model 2 denied bail		Model 3 Ln bail amount
	b	(SE)	b	(SE)	b	(SE)
Male defendant	−0.431	(0.046)***	0.544	(0.088)***	0.310	(0.023)***
Defendant age	0.005	(0.002)**	−0.005	(0.003)	−0.001	(0.001)
Defendant Black non-Hispanic	0.007	(0.047)	0.119	(0.070)	0.096	(0.021)***
Defendant Hispanic	−0.209	(0.052)***	0.157	(0.078)*	0.134	(0.025)***
Most serious charge-violent	−0.537	(0.074)***	0.376	(0.093)***	0.822	(0.031)***
Most serious charge-property	0.394	(0.068)***	−0.361	(0.095)***	−0.074	(0.030)*
Most serious charge-drug	0.172	(0.068)*	−0.428	(0.094)***	0.079	(0.030)**
Total number of charges	−0.122	(0.013)***	0.053	(0.017)**	0.180	(0.006)***
Prior felony arrest	−0.054	(0.009)***	0.053	(0.011)***	0.011	(0.004)**
Prior felony conviction	−0.106	(0.015)***	0.057	(0.016)***	0.058	(0.006)***
Active CLS status	−0.434	(0.048)***	1.283	(0.061)***	0.119	(0.020)***
Prior FTA	−0.019	(0.049)	0.042	(0.066)	0.028	(0.022)
Year = 2000	0.163	(0.069)*	−0.676	(0.081)***	0.449	(0.030)***
Year = 2009	0.002	(0.069)	−0.959	(0.085)***	0.532	(0.031)***
Constant	−1.627	(0.275)***	−4.328	(0.252)***	7.099	(0.131)***
N	31,974		31,974		21,173

Note: *p < 0.05; **p < 0.01; ***p < 0.001 (two-tailed).

Defendants charged with violent crimes are significantly less likely to receive ROR, more likely to be denied bail, and have significantly higher average bail amounts relative to those facing public order crimes; while those with property or drug charges are more likely to receive ROR and less likely to be denied bail. However, compared to public order offenses, those facing charges for property offenses receive lower bail amounts, while those facing drug charges receive higher bail amounts. For each additional charge being faced by a defendant, the odds of being given ROR decrease by 12%, while the odds of bail denial increase by 5% and bond amounts increase by 18%. Further, each prior felony arrest and conviction is associated with a decrease in the odds of ROR by five and 11%, respectively, and an increase in the odds of bail denial by 5% and 6%, respectively. While each prior felony arrest is associated with a 1% increase in a defendant's bond amount, each prior felony conviction is associated with a 6% increase in bond amount. Defendants with active criminal legal system statuses are 43% less likely to receive ROR, and over three and a half times more likely to be denied bail than those without active statuses. Moreover, they receive bond amounts that are 12% higher than those without active statuses. Interestingly, prior failures to appear do not influence pretrial detention decisions in our models, suggesting that “danger of flight” may not be as important to judges as the perceived dangerousness of a defendant.

Pretrial decisions vary by year. Defendants were significantly more likely to receive ROR in 2000 than in 1990; however, 2009 is not significantly different from 1990. Defendants are also significantly less likely to have bail denied in 2000 and 2009, relative to 1990, and the trend appears to be moving towards reduced use of bail denial, as the odds ratio drops dramatically from .51 in 2000 to .38 in 2009. Finally, relative to 1990, logged bail amounts are significantly higher in 2000 and 2009, suggesting the average logged bail across counties has increased over time, even once accounting for inflation.

County-level predictors of release on own recognizance

Table 3 presents the results of the full models predicting ROR. Model 4 indicates that county-level violent crime and economic inequality are associated with lower odds of ROR. Unemployment rates are positively associated with the likelihood of ROR. Further, percentage Black has a negative and significant impact on the odds of ROR, but this relationship is non-linear—once percentage Black reaches a threshold of approximately 30%, the relationship reverses and significantly increases the odds of ROR. Conversely, Black-to-White residential segregation is not significantly associated with ROR decisions.

Table 3.

Hierarchical logistic and linear regression estimates predicting the effect of county-level race and ethnicity measures on release on recognizance, denied bail, and bail amount.

	Release on recognizance (N = 31,974)				Denied bail (N = 31,974)				Bail amount (N = 21,173)
	Model 4 (Black)		Model 5 (Hispanic)		Model 6 (Black)		Model 7 (Hispanic)		Model 8 (Black)		Model 9 (Hispanic)
Violent crime rate	−0.090	(0.015)***	−0.099	(0.019)***	0.116	(0.017)***	0.082	(0.019)***	−0.077	(0.006)***	−0.053	(0.008)***
Unemployment rate	0.242	(0.040)***	0.236	(0.037)***	−0.034	(0.045)	−0.081	(0.040)*	0.048	(0.013)***	0.103	(0.012)***
Gini index	−4.669	(0.868)***	−5.033	(0.803)***	3.417	(1.041)**	3.977	(0.952)***	−0.309	(0.251)	0.625	(0.237)**
Percentage Black	−0.077	(0.030)**	—	—	0.022	(0.037)	—	—	−0.089	(0.011)***	—	—
Percentage Black squared	0.001	(0.000)**	—	—	−0.001	(0.001)	—	—	0.001	(0.000)***	—	—
BWSeg	−0.029	(0.038)	—	—	0.108	(0.056)	—	—	0.042	(0.018)*	—	—
BWSeg squared	0.001	(0.000)	—	—	−0.000	(0.000)	—	—	−0.001	(0.000)***	—	—
Percentage Hispanic	—	—	−0.036	(0.022)	—	—	0.051	(0.021)*	—	—	−0.014	(0.008)
Percentage Hispanic Squared	—	—	0.000	(0.000)	—	—	−0.001	(0.000)*	—	—	0.001	(0.000)***
HWSeg	—	—	0.118	(0.042)**	—	—	0.371	(0.070)***	—	—	−0.051	(0.019)**
HWSeg squared	—	—	−0.002	(0.001)**	—	—	−0.005	(0.001)***	—	—	0.001	(0.000)***
Year = 2000	0.081	(0.166)	−0.138	(0.248)	0.545	(0.215)*	−0.369	(0.263)	0.457	(0.074)***	0.029	(0.092)
Year = 2009	−1.285	(0.345)***	−1.699	(0.356)***	1.057	(0.345)**	−0.547	(0.307)	0.166	(0.114)	−0.506	(0.113)***
_Constant	−0.657	(1.199)	−1.895	(0.867)*	−13.470	(1.987)***	−12.661	(1.398)***	8.072	(0.553)***	7.565	(0.410)***

Note: The (—) em-dash indicates entries that are not applicable or omitted.

b: coefficient; SE: standard error; BWseg: Black-to-White residential segregation; HWSeg: Hispanic-to-White residential segregation.

Level 1 variables are included in the models, but coefficients and standard errors are suppressed.

*p < 0.05; **p < 0.01; ***p < 0.001 (two-tailed).

Model 5 presents the results of ethnic composition and segregation on ROR decisions. In line with Model 4, we find the significant positive effects for violent crime, unemployment, and economic inequality. However, we find no evidence of a relationship between percentage Hispanic and ROR outcomes. Hispanic-to-White residential segregation has a significant, non-linear effect on ROR. Hispanic-to-White residential segregation increases the odds of ROR until a dissimilarity score of approximately .40, after which the odds of ROR begin to decrease. Lastly, when controlling for county-level factors in both models, ROR is significantly less likely to be used by courts in 2009, relative to 1990.

County-level predictors of bail denial

Models 6 and 7 present the results of the multi-level logistic regression models predicting bail denial. Both models suggest that violent crime and economic inequality are significantly associated with higher odds of being denied bail. In Model 6, unemployment, racial composition, and racial segregation do not significantly impact the odds of defendants being denied bail. However, when accounting for racial composition and segregation, defendants are significantly more likely to be denied bail in 2000 and 2009, relative to 1990.

The results presented in Table 3, Model 7 indicate that percentage Hispanic and Hispanic-to-White segregation are both significantly associated with bail denial across counties. Increases in the percentage of Hispanic populations increases the odds that bail will be denied until a tipping point is reached, after which the relationship reverses. This pattern is consistent for county-level Hispanic residential segregation. Both results suggest that the odds of bail denial are lower at the highest and lowest levels of Hispanic composition and segregation, and peak when percentage Hispanic and the Hispanic-to-White segregation measures reach, on average, around 40%. When accounting for Hispanic composition and segregation, there is no significant difference in bail denial in 2000 or 2009, relative to 1990.

County-level predictors of bail amount

The results of the multi-level linear regression models predicting logged bail amount are presented in Models 8 and 9. Model 8 provides the estimates of logged bail amount regressed on racial composition and segregation. These results suggest that increases in county-level percentage Black are associated with lower bail amounts, until a tipping point is reached, around approximately 44%, at which point the relationship reverses, and average bail amounts begin to increase (see Figure 1(a)). However, the results related to racial segregation reveal a different trend. As illustrated in Figure 1(b), average logged bail amounts increase as segregation increases, until a tipping point is met (approximately 40%), after which average bail amounts begin to decrease.

Estimates from the models predicting logged bail amounts when accounting for ethnic composition and segregation are provided in Model 9. The results suggest that counties with the largest Hispanic populations exhibit the highest bail amounts (see Figure 1(c)). The trend for ethnic segregation is similar, although the exponential growth is not as pronounced. As shown in Figure 1(d), very low levels of ethnic segregation are associated with lower average bail amounts, until ethnic segregation reaches a dissimilarity index of around 30, wherein average bail amounts begin to significantly increase.

Discussion and conclusion

This study examined how indicators of county racial composition and segregation affect three important pretrial detention decisions. Our findings are nuanced. The relationship between percentage Black and ROR is curvilinear, resembling a U-shaped distribution. When counties have small Black populations, ROR tends to be high, and as percentage Black increases, the odds of ROR begin to decline. These declines continue until the Black and White populations start to reach parity, around 40%, and then the use of ROR begins to climb again. We interpret these findings as aligning with the key tenets of group threat theory—judges are more likely to release defendants under conditions of low ‘threat’—in this case, when the Black population is small and when it is sufficiently large enough to have accumulated power. When threat is most pronounced, that is, when Black populations are large, but not too large, the use of ROR is lowest.

The relationship between bail amount and racial composition follows a slightly different pattern. As percentage Black increases, average bail amounts decrease. Departing from the ROR findings, we find the association reverses direction once percentage Black reaches a tipping point. Of course, bail amount is contingent on earlier decision points (e.g. only defendants who haven’t been released or denied bail are required to pay), so these findings might reflect the reality that more defendants are released on their own recognizance in places with large Black populations, thereby artificially inflating the bail amounts paid by those who are not.

In line with Wang and Mears (2015) and recent research on minority neglect in incarceration (e.g. Beckman and Wang, 2022; Keen and Jacobs, 2009; Williams and Vaughn, 2024), we find support for the notion that criminal justice actors may punish less when county minority composition reaches a tipping point, which results in lower perceived threat and diminished social control (see also Liska and Chamlin, 1984). Average bail amounts are lowest at the lower and upper limits of racial segregation. Defendants appearing in court in counties with low segregation between White and Black residents tend to have lower average bail amounts. As Black-to-White segregation increases, average bail amounts increase and continue to do so until the Black and White segregation level hits approximately 40%, after which point bail amounts steadily decline. In line with group threat theory, when segregation—and social isolation between groups—is high enough, fear of crime and resource scarcity may be of less concern to criminal justice actors, who are likely responding to public opinion (e.g. Boyd and Nelson, 2017; Nelson, 2014; Pickett, 2019).

The findings regarding ethnic composition and punishment differ in key ways that also support the group threat perspective. In line with most research suggesting that formal social control strategies are used by criminal legal system actors to maintain the status quo (Meyer, 2000; Valenty and Sylvia, 2004), we find that percentage Hispanic is positively associated with average bail amount (but see Wang and Mears, 2015). However, and in contrast to recent work (e.g. Beckman and Wang, 2022; Zvonkovich and Ulmer, 2023), we find average bail amounts to be higher in counties with more Hispanic segregation. We suspect that while segregation by race is long-standing in large urban communities, the growing changes in Hispanic composition may lead to increased fear, as Hispanics become more spatially dispersed in large cities (Leach and Bean, 2008; Lichter et al., 2010).

Overall, the contrasting relationships found between Black and Hispanic population size and segregation suggest that punishment responses in America are largely racialized and ethnicized (Eitle and Taylor, 2008; Lehmann and Meldrum, 2024; Wang and Mears, 2015). Moreover, the significant, and at times conflicting, results regarding the impact of minority composition and residential segregation on bail or release decisions underscore the importance of continuing to examine both measures and their differential effects on county-level case-processing.

In addition to gaining insight into the unique experiences of defendants situated in areas characterized by varying degrees of minority composition and residential integration, we find that community factors such as socioeconomic disadvantage, unemployment, and violent crime affect pretrial detention decision-making in important and consequential ways. For example, violent crime rates influence decisions to detain pretrial comports with prior research. Moreover, the negative association between violent crime and bail amount, taken alone, might appear to go against key tenets of group threat theory, which portends greater threat perceptions to accompany high levels of violence and higher bail to ease heightened fear of victimization. However, when one considers that bail amounts are conditional on earlier decisions of whether to ROR and deny bail, these findings make more sense, for it is likely that bail amount only meaningfully varies when defendants are unlikely to be given ROR or denied bail.

We would be remiss not to acknowledge a handful of limitations that potentially undermine the results reported here. First, we are unable to account for variation in state statutes governing bail decisions and risk assessment. We are also unable to assess system constraints that may influence county-level variation in outcomes. Factors such as jail overcrowding, fiscal concerns, and judicial caseloads can place pressure on judges to release more defendants on their own recognizance and/or choose lower bail amounts to alleviate constraints (e.g. Steffensmeier et al., 1998). For instance, Williams (2016) found that judges in Florida are increasingly concerned with detaining defendants pretrial when jail capacities are lower. Future research should examine between-state and between-county variation in pretrial decision-making to determine how different bail-setting guidelines, as well as various jurisdictional constraints and pressures within these counties, operate to affect court actor decision-making. Further, we are unable to account for the historical treatment of various racial and ethnic minority populations across counties. Future research should attend to the unique historical experiences of different racial and ethnic populations as they relate to pretrial punitiveness. Given the well-documented impact that pretrial detention has on subsequent decisions, particularly at the sentencing phase (St Louis, 2024), future research should also explore the mediating role that pretrial detention plays in the relationship between county-level context and sentencing decisions. Finally, we acknowledge that our models rely on data from over 30,000 individuals. While we are confident that our findings align with our theoretical expectations, the curvilinear relationships reported here could be a function of our sample size—statistically significant but substantively small in magnitude.

Despite these limitations, it is clear that community context plays an important role in shaping individual-level punitive attitudes and perpetuation of the racialized social order (Meyer, 2000; Taylor, 1998). Counties vary in their punishing of defendants. This variation affects whether defendants are able to post bail and/or secure their release prior to trial. While the literature has continuously shown that defendant socioeconomic status can influence criminal legal system treatment, even when attempts are made to lower court actor discretion (e.g. Skeem et al., 2020), the current study suggests that aggregate factors may also be operating to perpetuate racial and ethnic disparities. Specifically, bail and pretrial detention may be used under certain circumstances to reinforce the racialized social order and the “status quo” of majority White groups. However, as the results suggest, the ways in which community racial and ethnic composition are measured have important theoretical implications. A more nuanced understanding of how race and ethnicity play a role in punishment decisions via the group threat perspective requires scholars to explore not only the relative size of minority populations in a community but also the extent to which the minority and majority group members are segregated from one another. Therefore, it is vital that studies continue to consider the relationships between community context, in particular the contrasting influence of ethno-racial composition and segregation, and criminal legal system treatment.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Joshua H Williams

Notes

Joshua H Williams is an Assistant Professor in the Department of Justice, Culture, and Social Change at the University of Minnesota Duluth. His research interests include pretrial detention, bail, sentencing, and politics and punishment.

Paige E Vaughn is an Assistant Professor of Sociology and Criminology at Spring Hill College in Mobile, AL. Her research interests include criminal justice punishment, violence, and race and inequality.

Matt Vogel is an Associate Professor in the School of Criminal Justice at the University at Albany, SUNY and a faculty affiliate in the Department of Sociology and the Center for Social and Demographic Analysis. His research focuses on the causes, consequences, and prevention of community violence and the role of population dynamics in explaining long term trends in crime and justice.

References

Aaltonen

Macdonald

Martikainen

, et al. (2013) Examining the generality of the unemployment–crime association. Criminology 51(3): 561–594.

Allport

(1954) The Nature of Prejudice. Cambridge, MA: Addison-Wesley.

Baumer

(2013) Reassessing and redirecting research on race and sentencing. Justice Quarterly 30(2): 231–261.

Baumer

Martin

(2013) Social organization, collective sentiment, and legal sanctions in murder cases. American Journal of Sociology 119(1): 131–182. The University of Chicago Press.

Beckman

Wang

(2022) Revisiting the minority threat perspective: Examining the main and interactive effects of segregation on sentencing severity. Justice Quarterly 39(4): 745–771.

Blalock

(1967) Toward a Theory of Minority-Group Relations. New York, NY: John Wiley and Sons. Available at: https://cir.nii.ac.jp/crid/1130000793701086720 (accessed 30 October 2023).

Blumer

(1958) Race prejudice as a sense of group position. Pacific Sociological Review 1(1): 3–7.

Bobo

Hutchings

(1996) Perceptions of racial group competition: Extending Blumer’s theory of group position to a multiracial social context. American Sociological Review 61(6): 951–972.

Boyd

Nelson

(2017) The effects of trial judge gender and public opinion on criminal sentencing decisions. Vanderbilt Law Review 70: 1819–1843.

10.

Britt

(2000) Social context and racial disparities in punishment decisions. Justice Quarterly 17(4): 707–732.

11.

Casillas

Enns

Wohlfarth

(2011) How public opinion constrains the US Supreme Court. American Journal of Political Science 55(1): 74–88.

12.

Chambliss

Seidman

(1971) Law, Order, and Power. Reading, MA: Addison-Wesley. Available at: https://cir.nii.ac.jp/crid/1130282272300148096 (accessed 25 January 2024).

13.

Charles

(2003) The dynamics of racial residential segregation. Annual Review of Sociology 29(1): 167–207.

14.

Chiricos

McEntire

Gertz

(2001) Perceived racial and ethnic composition of neighborhood and perceived risk of crime. Social Problems 48(3): 322–340.

15.

Davis

(2007) Arbitrary Justice: The Power of the American Prosecutor. New York, NY: Oxford University Press.

16.

Demuth

(2003) Racial and ethnic differences in pretrial release decisions and outcomes: A comparison of hispanic, black, and white felony arrestees. Criminology 41(3): 873–908.

17.

Dhami

(2005) From discretion to disagreement: Explaining disparities in judges’ pretrial decisions. Behavioral Sciences & the Law 23(3): 367–386.

18.

Donnelly

(2022) Neighborhoods, criminal incidents, race, and sentencing: Exploring the racial and social context of disparities in incarceration sentences. The British Journal of Criminology 62(1): 145–164.

19.

Donnelly

Asiedu

(2020) Neighborhood context and racial/ethnic disparities in judicial processing. Sociology Compass 14(9): e12822.

20.

Duxbury

(2021) Who controls criminal law? Racial threat and the adoption of state sentencing law, 1975 to 2012. American Sociological Review 86(1): 123–153.

21.

Eitle

D’Alessio

Stolzenberg

(2002) Racial threat and social control: a test of the political, economic, and threat of black crime hypotheses. Social Forces 81(2): 557–576.

22.

Eitle

Taylor

(2008) Are hispanics the new ‘threat’? Minority group threat and fear of crime in Miami-dade county. Social Science Research 37(4): 1102–1115.

23.

Fearn

(2005) A multilevel analysis of community effects on criminal sentencing. Justice Quarterly 22(4): 452–487.

24.

Feldmeyer

Cochran

(2018) Racial threat and social control: A review and conceptual framework for advancing racial threat theory. In: Unnever

Gabbidon

Chouhy

(eds) Building a Black Criminology, Volume 24. New York, NY: Routledge, 283–316.

25.

Forman

Jr (2017) Locking up Our Own: Crime and Punishment in Black America. New York, NY: Farrar, Straus and Giroux.

26.

Fox

(2015) Applied Regression Analysis and Generalized Linear Models. Thousand Oaks, CA: Sage Publications.

27.

Freiburger

Hilinski

(2010) The impact of race, gender, and age on the pretrial decision. Criminal Justice Review 35(3): 318–334.

28.

Freiburger

Marcum

Pierce

(2010) The impact of race on the pretrial decision. American Journal of Criminal Justice 35(1–2): 76–86.

29.

Goldkamp

Gottfredson

(1979) Bail decision making and pretrial detention. Law and Human Behavior 3(4): 227–249.

30.

Greenberg

Kessler

Loftin

(1985) Social inequality and crime control. The Journal of Criminal Law and Criminology (1973-) 76(3): 684.

31.

Greenberg

West

(2001) State prison populations and their growth, 1971–1991. Criminology 39(3): 615–654.

32.

Helms

Jacobs

(2002) The political context of sentencing: An analysis of community and individual determinants. Social Forces 81(2): 577–604.

33.

Hood

Schneider

(2019) Bail and pretrial detention: Contours and causes of temporal and county variation. RSF: The Russell Sage Foundation Journal of the Social Sciences 5(1): 126–149.

34.

Huber

Gordon

(2004) Accountability and coercion: Is justice blind when it runs for office? American Journal of Political Science 48(2): 247–263.

35.

Johnson

(2006) The multilevel context of criminal sentencing: Integrating judge-and county-level influences. Criminology 44(2): 259–298.

36.

Keen

Jacobs

(2009) Racial threat, partisan politics, and racial disparities in prison admissions: A panel analysis. Criminology 47(1): 209–238.

37.

King

Light

(2019) Have racial and ethnic disparities in sentencing declined? Crime and Justice 48(1): 365–437.

38.

Klein

Wittes

(2011) Preventive detention in American theory and practice. Harvard National Security Journal 2: 85–191.

39.

Kramer

Ulmer

(2002) Downward departures for serious violent offenders: Local court “corrections” to Pennsylvania’s sentencing guidelines. Criminology 40(4): 897–932.

40.

Krivo

Peterson

Kuhl

(2009) Segregation, racial structure, and neighborhood violent crime. American Journal of Sociology 114(6): 1765–1802.

41.

Kurlychek

Johnson

(2019) Cumulative disadvantage in the American criminal justice system. Annual Review of Criminology 2(1): 291–319.

42.

Leach

Bean

(2008) The structure and dynamics of Mexican migration to new destinations in the United States. In: Massey

(eds) New Faces in New Places: The Changing Geography of American immigration. New York, NY: Russell Sage Foundation, 51–74.

43.

Legewie

Schaeffer

(2016) Contested boundaries: Explaining where ethnoracial diversity provokes neighborhood conflict. American Journal of Sociology 122(1): 125–161.

44.

Lehmann

Meldrum

(2024) Disparities in youth arrest across racial and ethnic subgroups. Youth Violence and Juvenile Justice 22(1): 22–45.

45.

Leonard

(1987) The interaction of residential segregation and employment discrimination. Journal of Urban Economics 21(3): 323–346.

46.

Lichter

Parisi

Taquino

, et al. (2010) Residential segregation in new hispanic destinations: Cities, suburbs, and rural communities compared. Social Science Research 39(2): 215–230.

47.

Liska

(1992) Social Threat and Social Control. Albany, NY: SUNY Press.

48.

Liska

Chamlin

(1984) Social structure and crime control among macrosocial units. American Journal of Sociology 90(2): 383–395.

49.

Lynch

(2011) Mass incarceration, legal change, and locale: Understanding and remediating American penal overindulgence. Criminology & Public Policy 10(3): 673–698.

50.

Lyons

(2007) Community (dis) organization and racially motivated crime. American Journal of Sociology 113(3): 815–863.

51.

Markert

(2010) The changing face of racial discrimination: Hispanics as the dominant minority in the USA- A new application of power-threat theory. Critical Sociology 36(2): 307–327.

52.

Massey

Condran

Denton

(1987) The effect of residential segregation on black social and economic well-being. Social Forces 66(1): 29–56.

53.

Massey

Denton

(1993) American Apartheid: Segregation and the Making of the Underclass. Cambridge, MA: Harvard University Press.

54.

Meyer

(2000) As Long as They Don’t Move next Door: Segregation and Racial Conflict in American Neighborhoods. Lanham, MD: Rowman & Littlefield.

55.

Nelson

(2014) Responsive justice? Retention elections, prosecutors, and public opinion. Journal of Law and Courts 2: 117–152.

56.

Nicholson-Crotty

(2004) The impact of sentencing guidelines on state-level sanctions: An analysis over time. Crime & Delinquency 50(3): 395–411.

57.

Ouellette

Applegate

(2023) Local incarceration as social control: A national analysis of social, economic, and political determinants of jail use in the United States. American Journal of Criminal Justice 48(3): 547–571.

58.

Parker

Stults

Rice

(2005) Racial threat, concentrated disadvantage and social control: Considering the macro-level sources of variation in arrests. Criminology 43(4): 1111–1134.

59.

Peterson

Krivo

(1993) Racial segregation and black urban homicide. Social Forces 71(4): 1001–1026.

60.

Pettigrew

Tropp

(2006) A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology 90(5): 751.

61.

Piatkowska

Messner

(2022) Group threat, same-sex marriage, and hate crime based on sexual orientation. Justice Quarterly 39(4): 802–824.

62.

Pickett

(2019) Public opinion and criminal justice policy: Theory and research. Annual Review of Criminology 2(1): 405–428.

63.

Pinchevsky

Steiner

(2016) Sex-based disparities in pretrial release decisions and outcomes. Crime & Delinquency 62(3): 308–340.

64.

Reaves

(2013) Felony Defendants in Large Urban Counties, 2009 Statistical Tables. Washington, D.C.: US Department of Justice, Bureau of Justice Statistics.

65.

Rothstein

(2017) The Color of Law: A Forgotten History of How Our Government Segregated America. New York, NY: Liveright Publishing.

66.

Sajaia

(2007) FASTGINI: Stata Module to Calculate Gini Coefficient with Jackknife Standard Errors. Chestnut Hill, MA: Statistical Software Components. Available at: https://econpapers.repec.org/software/bocbocode/s456814.htm (accessed 25 January 2024).

67.

Schlesinger

(2005) Racial and ethnic disparity in pretrial criminal processing. Justice Quarterly 22(2): 170–192.

68.

Shihadeh

Flynn

(1996) Segregation and crime: The effect of black social isolation on the rates of black urban violence. Social Forces 74(4): 1325–1352.

69.

Skeem

Scurich

Monahan

(2020) Impact of risk assessment on judges’ fairness in sentencing relatively poor defendants. Law and Human Behavior 44(1): 51–59.

70.

Spitzer

(1975) Toward a Marxian theory of deviance. Social Problems 22(5): 638–651.

71.

Steffensmeier

Ulmer

Kramer

(1998) The interaction of race, gender, and age in criminal sentencing: The punishment cost of being young, black, and male. Criminology 36(4): 763–798.

72.

St. Louis

(2024) The pretrial detention penalty: A systematic review and meta-analysis of pretrial detention and case outcomes. Justice Quarterly 41(3): 347–370.

73.

Stolzenberg

D’Alessio

Eitle

(2013) Race and cumulative discrimination in the prosecution of criminal defendants. Race and Justice 3(4): 275–299.

74.

Stults

Baumer

(2007) Racial context and police force size: Evaluating the empirical validity of the minority threat perspective. American Journal of Sociology 113(2): 507–546.

75.

Sutton

(2013) Structural bias in the sentencing of felony defendants. Social Science Research 42(5): 1207–1221.

76.

Taylor

(1998) How white attitudes vary with the racial composition of local populations: Numbers count. American Sociological Review 63(4): 512–535.

77.

Ulmer

(2012) Recent developments and new directions in sentencing research. Justice Quarterly 29(1): 1–40.

78.

Ulmer

Johnson

(2004) Sentencing in context: A multilevel analysis. Criminology 42(1): 137–178.

79.

Ulmer

Kurlychek

Kramer

(2007) Prosecutorial discretion and the imposition of mandatory minimum sentences. Journal of Research in Crime and Delinquency 44(4): 427–458.

80.

Valenty

Sylvia

(2004) Thresholds for tolerance: The impact of racial and ethnic population composition on the vote for California propositions 187 and 209. The Social Science Journal 41(3): 433–446.

81.

Vogel

Lentz

Steenbeek

(2023) Contested neighborhood boundaries and intergroup and intragroup violence. Crime & Delinquency 70(11): 2906–2931.

82.

Vogel

Messner

(2024) Group threat and social control: Who, what, where, and when. Annual Review of Criminology 7(1): 39–58.

83.

Wacquant

(2001) Deadly symbiosis: When ghetto and prison meet and mesh. Punishment & Society 3(1): 95–133.

84.

Wang

Mears

(2010) A multilevel test of minority threat effects on sentencing. Journal of Quantitative Criminology 26(2): 191–215.

85.

Wang

Mears

(2015) Sentencing and state-level racial and ethnic contexts. Law & Society Review 49(4): 883–915.

86.

Weidner

Frase

(2003) Legal and extralegal determinants of intercounty differences in prison use. Criminal Justice Policy Review 14(3): 377–400.

87.

Weidner

Frase

Schultz

(2005) The impact of contextual factors on the decision to imprison in large urban jurisdictions: A multilevel analysis. Crime & Delinquency 51(3): 400–424.

88.

Welch

Payne

Chiricos

, et al. (2011) The typification of hispanics as criminals and support for punitive crime control policies. Social Science Research 40(3): 822–840.

89.

Williams

Collins

(2001) Racial residential segregation: A fundamental cause of racial disparities in health. Public Health Reports 116(5): 404–416.

90.

Williams

Campbell

(2021) Exploring the time-varying determinants of state spending on corrections. Journal of Quantitative Criminology 37(3): 671–692.

91.

Williams

Rosenfeld

(2016) The impact of neighborhood status on imprisonment for firearm offenses. Journal of Contemporary Criminal Justice 32(4): 383–400.

92.

Williams

Vaughn

(2024) Cheap on punishment: Examining the impact of prison population racial demographics on state-level corrections spending. Justice Quarterly 41(4): 597–618.

93.

Williams

(2016) From bail to jail: The effect of jail capacity on bail decisions. American Journal of Criminal Justice 41(3): 484–497.

94.

Wooldredge

(2007) Neighborhood effects on felony sentencing. Journal of Research in Crime and Delinquency 44(2): 238–263.

95.

Wooldredge

(2012) Distinguishing race effects on pre-trial release and sentencing decisions. Justice Quarterly 29(1): 41–75.

96.

Wooldredge

Frank

Goulette

, et al. (2015) Is the impact of cumulative disadvantage on sentencing greater for black defendants? Criminology & Public Policy 14(2): 187–223.

97.

Zvonkovich

Ulmer

(2023) Segregation and group threat: Specifying Hispanic-white punishment disparity. Social Problems spad 032: 1–21.