Ecological Contributors to Disparities in Bond Amounts and Pretrial Detention

Abstract

Pretrial dispositions have been receiving greater attention in the literature on extralegal disparities in criminal case processing. We examined the relevance of areas in which crimes are committed for court decisions regarding bond amounts and whether suspects are ultimately detained prior to trial. A random sample of 2,677 persons charged with felony crimes committed in 820 blocks of a major urban U.S. jurisdiction was examined, with separate analyses of property, violent, and drug offenses. Defendants were more likely to be held in jail prior to trial when crimes were committed in more disadvantaged neighborhoods (higher percentages of female-headed households, vacant residences, renters, and African Americans). However, the odds of pretrial detention were also higher for defendants accused of crimes in less disadvantaged neighborhoods relative to their own. Evidence favors neighborhood composition as an important contributor to disparities in pretrial detention beyond individual factors such as a defendant’s race.

Keywords

disparities in case processing bond amount pretrial detention neighborhood disadvantage

Introduction

The empirical literature on extralegal disparities in criminal case processing is most often grounded in critical frameworks related to how the criminal justice system operates to favor defendants with higher socioeconomic status (SES) at the expense of lower status suspects because the latter are identified as more “dangerous” (Albonetti, 1991; Black, 1976; Hawkins, 1981; Steffensmeier, Ulmer, & Kramer, 1998) and/or as a threat to the interests of power holders (Blalock, 1967; Blumer, 1955; Bobo & Hutchings, 1996). In the criminal courts, therefore, lower SES defendants may experience more punitive treatment relative to higher SES defendants because of their marginalized status (Reiman, 2001). With few exceptions, court scholars have focused primarily on individual-level indicators of social disadvantage in related research. Recent movements by prosecutors to crack down on crimes in more disadvantaged neighborhoods in pursuit of “community justice” (Karp & Clear, 2000), however, have generated a macro-level relevance to these theories. Cracking down on crimes in poor areas necessarily absorbs more socially disadvantaged persons into the criminal justice system, potentially adding yet another dimension to the disparity literature. Also relevant to a neighborhood focus, and consistent with discussions of how the status differential between an offender and a victim might shape treatment by the system (Black, 1976; LaFree, 1980; Spohn & Spears, 1996), offenders from lower status neighborhoods who travel to higher status areas to commit their crimes might be treated even more harshly than those who commit crimes in their own or similar neighborhoods.

There is a growing body of empirical findings suggesting that pretrial detention poses a significant disadvantage to defendants by increasing their odds of conviction and imprisonment (a review of this literature plus contrary findings is provided by Reitler, Sullivan, & Frank, 2013; see also Sutton, 2013), even aside from the harsher treatment experienced by the nature of pretrial confinement. This body of research has generated more interest in pretrial dispositions and legal versus extralegal effects on bond amounts and pretrial detention. In light of court actors’ concerns with community justice, we examined whether defendants charged with committing crimes in more disadvantaged neighborhoods were assigned higher bond amounts and faced higher odds of pretrial detention, and whether status differentials between these neighborhoods and defendants’ residential neighborhoods also mattered for shaping these outcomes. A random sample of 2,677 persons charged with felony crimes committed in 820 blocks of a major urban U.S. jurisdiction was examined, with property, violent, and drug offenses examined separately.

Framework

Critical perspectives on extralegal disparities in case processing are generally framed in symbolic interactionism at the individual level, arguing that court actors may “type” defendants as more culpable or at higher risk for future offending when they possess certain attributes that reinforce their images of more dangerous offenders, such as being African American, young, male, unemployed, and so forth (e.g., Nobiling, Spohn, & DeLone, 1998; Spohn & Holleran, 2000; Steffensmeier et al., 1998). These defendants may face harsher treatment at each stage of case processing, from indictment through sentencing, if judges and prosecutors base their decisions in part on these extralegal factors (Kutateladze, Andiloro, Johnson, & Spohn, 2014). This process of “perceptual shorthand” (Hawkins, 1981, p. 230) serves to expedite case processing, consistent with Albonetti’s (1987, 1991) “uncertainty avoidance” perspective on how judges and prosecutors attempt to reduce uncertainty in their decisions through consideration of what they perceive as the more typical attributes of higher risk offenders they have dealt with in the past. Such considerations are an important element of Steffensmeier et al.’s (1998) focal concerns theory, and the idea that court actors’ concerns with controlling and reducing crime might lead them to stereotype persons with certain characteristics as greater threats to public safety. Although most related theories were developed with sentencing practices in mind (but see Albonetti, 1987), these perspectives also apply to pretrial release decisions because those decisions are subject to the same abuses of discretion, as discussed below (also see Demuth’s, 2003, application of focal concerns to an understanding of pretrial dispositions).

Public safety is a concern of both judges and prosecutors at initial appearance, and court actors involved in recommending and determining bond amounts, denial of bond, and release on recognizance (ROR) might be influenced by stereotypes of more dangerous offenders to reduce the risk of public harm (Demuth, 2003). Suspects’ personal attributes such as race, sex, and/or age might be elements of these stereotypes, and the influence of such factors on pretrial detention might be compounded if these factors are also linked to the ability to post a given bond. For example, proportionately more minority defendants in more segregated cities might be held in jail prior to trial if they are less able to post bond (Richey-Mann, 1993). A suspect’s financial resources can be considered by judges when determining bond amounts (Goldkamp, 1979; Goldkamp & Gottfredson, 1985), but it is impossible for some to post any bond regardless of the amount (Foote, 1954).

Bond amounts are ultimately set by judges but, in the jurisdiction examined here, are also influenced by bond schedules and recommendations of prosecutors. Pretrial detention, by contrast, is primarily an outcome rather than a decision, with the exceptions of ROR and the denial of bond (also fed by input from prosecutors and judges’ final determinations). The vast majority of suspects are assigned some type of bond, however, and so the ability to post a given bond amount is out of the hands of court actors. Here, we focus on both bond amounts and pretrial detention, recognizing that analysis of the latter is not an analysis of a “decision” per se (Demuth, 2003). Nonetheless, it is ultimately influenced by court actors’ recommendations and decisions (ROR, denial of bond, and bond amount) and, most relevant to our interest, can be shaped by extralegal factors either in terms of how they influence the first two decisions or how they present barriers to meeting the third.

Empirical studies of bond amounts and pretrial release have not examined neighborhood effects on these dispositions, the relevance of which is discussed below. Research to date on extralegal disparities in pretrial dispositions has focused primarily on a defendant’s race and pretrial detention, with some findings of no direct race effects on pretrial detention when controlling for offense seriousness and prior record (e.g., Albonetti, 1989; Frazier, Bock, & Henretta, 1980; Holmes, Daudistel, & Farrell, 1987, 1996; Nagel, 1983; Stryker, Nagel, & Hagan, 1983; Wooldredge, 2012), and other findings of significant direct race effects and/or indirect race effects via bond amounts even with these controls (e.g., Ayres & Waldfogel, 1994; Chiricos & Bales, 1991; Demuth, 2003; Demuth & Steffensmeier, 2004; Katz & Spohn, 1995; LaFree, 1985; Lizotte, 1978; Wooldredge, Frank, Goulette, & Travis, 2015). Other scholars have also noted more favorable treatment of females relative to males (Daly, 1989; Demuth & Steffensmeier, 2004; Goulette, Wooldredge, Frank, & Travis, 2015; Maxwell, 1999), perhaps due to perceptions of women as less threatening to public safety and as the primary caregivers for their children. Significant age effects were also uncovered by Demuth (2003), who found that ROR was more common for suspects younger than 30 and those older than 50.

Specific to bond amounts, evidence seems to favor the idea that bail is driven primarily by the severity of charges and available evidence (Bock & Frazier, 1978; Goldkamp & Gottfredson, 1985; Goldkamp, Gottfredson, & Mitchell-Herzfeld, 1981; Gottfredson & Gottfredson, 1988; Nagel, 1983; Thomas, 1976; Walker, 2006), but this does not necessarily rule out the significance of extralegal factors because most of these studies focused primarily on whether and by how much bail is actually tied to legal considerations. For example, Wooldredge et al. (2015) uncovered significantly higher bond amounts for young African American men even though legal factors (charge severity, types of offense, specifications, and criminal history) were far more relevant predictors.

As recently discussed by Vîlcică and Goldkamp (2015), as is evident in our review above, the body of empirical research on pretrial dispositions has been “individual-focused” and has ignored the role of community context for shaping decisions (e.g., bail amounts) as well as events (e.g., failure to appear). Although they did not specifically examine bail amounts and pretrial detention, their work is relevant to ours as they considered the role of neighborhood context for shaping pretrial events (failure to appear and pretrial crime), thereby moving beyond the traditional focus on individual factors only. Using a sample of 800 defendants from 45 Philadelphia neighborhoods, Vîlcică and Goldkamp found that neighborhood SES was a significant predictor of pretrial crime. Not only does their general discussion of the importance of moving beyond individual-level models for a more complete understanding of pretrial events highlight the relatively unique contribution of our own analysis, but Vîlcică and Goldkamp’s significant finding for neighborhood SES underscores the relevance of the ensuing discussion of “neighborhood” as a focal concern of court actors in pretrial decisions.

“Neighborhood” as a Focal Concern of Judges and Prosecutors

From a focal concerns perspective, interest by judges and prosecutors in reducing threats to public safety in more crime-ridden areas of their cities leaves open the possibility that characteristics of the neighborhoods where crimes are committed may influence suspects’ pretrial dispositions. Given that the characteristics of neighborhoods in which crimes occur might play a role in police officers’ use of discretion and whether they arrest a suspect (Klinger, 1997), it is intuitive to expect that court actors might consider the same when assessing a suspect’s threat to safety in those neighborhoods. Scholars have found evidence to support the argument that police stops and aggressive tactics are disproportionately used with poor people in poor places (Fagan & Davies, 2000; Smith, 1986; Smith, Visher, & Davidson, 1984; Terrill & Reisig, 2003). Following this argument, suspects arrested for crimes in areas with greater social and economic disadvantage might be treated more harshly by court actors (higher bond amounts, and if they live in these areas, less eligible for ROR, or more likely to be denied bond altogether).

This discussion overlaps with how the pursuit of “community justice” might shape judicial and prosecutorial behaviors (Karp & Clear, 2000). Court actors may consider the quality of community life when processing defendants. The social context of the area in which a crime occurs may be a factor in decision making when crime and disadvantage are problems in those areas and legal authorities are consciously trying to “clean up” those neighborhoods. Pretrial dispositions might be harsher when defendants are accused of committing crimes in areas where crime itself is perceived as contributing to the deterioration of community life, all else being equal. Both judges and prosecutors may want to deter or remove from the streets offenders who are prone to committing crimes in these areas. Criminal opportunities are generally greater in neighborhoods with higher levels of economic disadvantage, more transient residents, more vacant structures, and so forth, and court actors may consider these attributes as “risk” factors to be considered in their decisions, similar to considerations of a defendant’s individual attributes that reinforce court actors’ images of more dangerous offenders. Vîlcică and Goldkamp’s (2015) finding of a significant inverse effect of neighborhood SES on a defendant’s odds of committing crime while on pretrial release enhances the credibility of the idea that court actors consider these types of neighborhood factors in pretrial decisions. An empirical link between pretrial “risk” and neighborhood status poses the conundrum of how to efficiently predict pretrial risk of flight or new crimes without considering community factors and contributing further to extralegal disparities in criminal processing.

Williams (2015) also discussed the relevance of macro-level influences on bail decisions from a focal concerns perspective although her macro-level factor was rated jail capacity (vs. neighborhood disadvantage in this study) framed within Steffensmeier et al.’s (1998) third focal concern regarding available resources and the implications of judicial decisions for the criminal justice system (vs. their second focal concern involving protection of the community). Even so, and similar to Vîlcică and Goldkamp’s (2015) study, her study stands out by moving beyond the traditional focus on individual-level effects on pretrial dispositions.

Also relevant to a discussion of neighborhood effects on court decisions is the idea that where offenders commit their crimes relative to where they live might also affect pretrial dispositions. Separate from a community justice perspective and more in line with “power threat” (Blalock, 1967), offenders who migrate outside their neighborhoods to offend in more affluent neighborhoods might be perceived by judges and prosecutors as greater threats to the social order because of their willingness to victimize persons of higher SES. When speaking of race relations and crime control, Black (1976) discussed how agents of the criminal justice system might view crimes by racial minorities against Whites as representing a particularly disturbing violation of social boundaries. Race was much more “class-based” when Black wrote, but his discussion applies to social disadvantage more broadly.

This perspective stands in contrast to the idea of community justice because the latter conveys a greater sense of concern by court actors over the welfare of residents of more disadvantaged areas. A power threat perspective, by contrast, suggests that offenders from lower status neighborhoods who travel to higher status areas to commit their crimes might be treated even more harshly than those who commit crimes in their own or similar neighborhoods. Stronger perceptions of “threat” may be met with stronger efforts to control these offenders. In short, larger status differentials in neighborhood disadvantage between the incident and a suspect’s residence may coincide with harsher treatment by the courts, including higher bond amounts and higher odds of pretrial detention (due to less use of ROR and more denial of bond).

Research Setting

The analysis described here focused on a portion of cases sampled for a broader study of felony case processing in a very large trial court of general jurisdiction located in the Northern United States. We cannot identify the county as a condition of the study although we can provide some general information that should help readers to place the findings in proper context. The county population includes roughly one-third African Americans. Relative to other urban counties in the United States, it ranks high in terms of racial segregation with over half of the population living in racially homogeneous zip code areas. Political and social tensions between African Americans and Whites are also evident in the city, particularly with regard to crime control, as revealed by editorials in the local newspaper and published comments by citizens serving on a police review board for the city. The study described here also focused on cases processed in 2009, at which time the city faced a very high home foreclosure rate. Even prior to the recession, fewer than 15% of census tracts in the city limits were gentrified. The combination of population composition, political and social conflict, and economic recession makes the jurisdiction under study the archetypical area for the potential applicability of power threat.

Whether arrested for a misdemeanor or a felony, the initial appearance (where bond is usually set) occurs in municipal court with the exception of the most severe grade of felonies which immediately move to the trial court of general jurisdiction. If not waived by the accused, the subsequent preliminary hearing (where bond is continued or might be set) also takes place in municipal court, again with the exception of the most severe felonies. Once a felony case is bound over, jurisdiction shifts from municipal court to the trial court of general jurisdiction. Judges assigned at arraignment are not part of the group who set bond, including those assigned to the most severe felony cases. The jurisdiction does not operate under bail guidelines. Bail amounts are informed by a combination of bond schedules and “usual amounts” (as described to us), state law and agency rules, and a risk assessment instrument. As explained to suspects, a judge sets the bond amount based on the severity of the arrest charge(s), the suspect’s criminal record, and the suspect’s community ties. In short, there appears to be some room for discretion in these decisions. Bail investigators are used to help make recommendations to judges, and they consult with attorneys to this end.

Method

Our analysis was designed to test these study hypotheses:

Hypothesis 1: Bond amounts are higher for persons charged with felony crimes in more socially disadvantaged neighborhoods.

Hypothesis 2: Bond amounts are higher for persons charged with felony crimes in less disadvantaged neighborhoods relative to their own.

Hypothesis 3: The odds of pretrial detention are higher for persons charged with felony crimes in more socially disadvantaged neighborhoods.

Hypothesis 4: The odds of pretrial detention are higher for persons charged with felony crimes in less disadvantaged neighborhoods relative to their own.

Sample

The County Prosecutors’ Office (CPO) provided access to information on all persons referred in 2009 for felony offenses. A simple random sample of 5,000 persons was selected from the population of 18,407 referrals. We were ultimately interested in a 22% sample (4,000 persons) but selected 5,000 individuals due to a very large portion of drug cases processed in this jurisdiction (roughly a quarter of all cases). So as not to overwhelm the analyses with drug offenses, we oversampled by 1,000 referrals and skipped every other drug suspect in the sample to yield a final pool of roughly 4,000 persons. Data were compiled only for the final pool and not for the “extra” drug cases produced with the sample design, based on available resources, so they were not included in the analysis described below.

The analysis required information on the address of each incident as well as the address of each defendant. Incident addresses were the primary problem in this regard because roughly one third were either incomplete or incorrect based on missing or unidentifiable street names and dwelling numbers that fell outside the actual range for a given street. These individuals had to be removed from the analysis because the addresses could not be geocoded. After also removing an additional 152 referrals due either to incomplete background information or to multiple appearances in the sample, in which case we chose the first referral in 2009, we were left with 2,677 persons (N₁) for the analysis, or a 15% sample of the 2009 population of referrals. Checks for nonrandomness of the “missing” addresses revealed no pattern to the missing data in that the distributions of the individual-level measures described in the next section were virtually identical between the nonmissing and missing cases.

Variables

Table 1 describes the measurement and sample distribution of each variable in the analysis. The univariate descriptive statistics are displayed for the pooled sample as well as for the three general offense groups examined (property, violent, and drug crimes). The sum of property, violent, and drug offenders is less than the number of “pooled” cases because the pooled sample also included offenders with multiple charges for two or all three of these offense types. Overlap between the separate offense groups was removed by only including persons charged with one of these three offense types. The models described below were estimated for the pooled sample as well as for each of the subsamples to assess the robustness of findings across offense groups.

Table 1.

Description of Variables for the Pooled Sample and Offense Subsamples.

Variables	Pooled	Property	Violent	Drugs
Variables	M (s)	M (s)	M (s)	M (s)
Level 1: Felony defendants
Outcomes
Bond amount log_e	8.78 (1.47)	8.62 (1.46)	10.14 (1.45)	8.44 (1.19)
Pretrial detention	0.30 (0.46)	0.31 (0.46)	0.48 (0.50)	0.15 (0.36)
Independent variables
Male	0.81 (0.39)	0.78 (0.41)	0.84 (0.36)	0.85 (0.36)
Age 18-29	0.50 (0.50)	0.50 (0.50)	0.52 (0.50)	0.55 (0.50)
Age 30-49	0.40 (0.49)	0.40 (0.49)	0.40 (0.49)	0.36 (0.48)
Age 50+ (reference)	0.10 (0.30)	0.10 (0.29)	0.08 (0.28)	0.09 (0.29)
African American	0.70 (0.46)	0.66 (0.47)	0.74 (0.44)	0.79 (0.41)
Most serious charge
Felony 1	0.17 (0.38)	0.18 (0.39)	0.56 (0.50)	0.05 (0.22)
Felony 2	0.17 (0.37)	0.20 (0.40)	0.30 (0.46)	0.08 (0.28)
Felony 3	0.16 (0.37)	0.11 (0.31)	0.04 (0.21)	0.26 (0.44)
Felony 4	0.16 (0.36)	0.14 (0.35)	0.07 (0.26)	0.25 (0.43)
Felony 5 (reference)	0.34 (0.47)	0.37 (0.48)	0.03 (0.17)	0.36 (0.48)
Total charges (square root)	1.68 (0.54)	1.69 (0.52)	1.89 (0.51)	2.09 (0.36)
Murder suspect	0.00 (0.04)	—	0.01 (0.09)	—
Gun specification	0.07 (0.25)	—	0.15 (0.36)	—
Repeat violent offender specification	0.02 (0.15)	—	0.11 (0.31)	—
Victim physically injured	0.17 (0.38)	—	0.74 (0.44)	—
No. of victims of violence	0.25 (0.58)	—	0.73 (0.79)	—
Prior prison term served	0.43 (0.50)	0.41 (0.49)	0.41 (0.49)	0.53 (0.50)
Status difference in level of disadvantage between incident street block versus suspect’s residential street block	0.05 (0.81)	0.19 (0.93)	−0.02 (0.69)	0.03 (0.77)
N ₁	2,677	974	627	330
Level 2: Street blocks
Neighborhood disadvantage (composite of four items; Cronbach’s α = .81)	0.00 (1.00)	0.31 (0.97)	0.57 (0.91)	0.65 (0.89)
Percent African American residents	43.1 (37.7)	42.8 (37.5)	51.7 (37.1)	53.4 (37.3)
Percent female-headed households with children	38.1 (18.2)	38.3 (18.4)	43.5 (17.1)	45.2 (17.1)
Percent vacant structures	15.5 (10.1)	15.7 (10.2)	17.9 (10.1)	18.6 (9.86)
Percent renters	45.6 (24.6)	47.6 (25.1)	51.8 (22.5)	54.1 (22.3)
N ₂	820	525	393	232

Note. (s)= standard deviation.

The dependent variables for the analysis included bond amount and pretrial detention. Bond amounts were transformed into the natural log (log_e) to remove the skew of the distribution (Fox, 2008). Bond amount was examined for bond-eligible defendants only as bonds are not set for individuals denied pretrial release and for those released on their own recognizance. The binary measure of pretrial detention was examined for all felony indictments and compared persons who were released outright (bond posted or released on their own recognizance) to those initially detained prior to trial (no bond posted or denied release). The “detained” group included persons detained for the entire pretrial period as well as those initially detained for some time period but subsequently released prior to trial. This definition fits with the idea that delays in initial release, even when release is subsequently obtained, would reflect disparate harm when similarly situated defendants obtain release outright for the same offenses. On the other hand, defendants who obtained release outright but were subsequently confined because of new crimes were still categorized as being released outright because it is the initial state of detention that is most relevant to the reactions of court actors to the case at hand and because subsequent revocations were influenced by unmeasured events transpiring after initial release.

All the individual-level (Level 1) independent variables were treated as statistical control variables for the analysis of neighborhood (Level 2) effects on pretrial dispositions. Some of these, such as a defendant’s race, are also substantively interesting from the standpoint of extralegal disparities in case processing, so we devote some attention to those findings as well.

The indicators of a defendant’s age reflect Steffensmeier et al.’s (1998) discussion of how convicted defendants were most likely to be sent to prison if they fell into the 18 to 29 age group, whereas those older than 50 were least likely to go to prison. Although we focus here on pretrial dispositions, the same logic might apply in that the youngest defendants might face the highest bond amounts and the highest odds of pretrial detention (on average), whereas the oldest defendants might face the lowest bond amounts and the lowest odds of detention. Others have also examined these age groups in their analyses of pretrial dispositions (e.g., Demuth, 2003; Demuth & Steffensmeier, 2004; Wooldredge, 2012). Similarly, there is empirical evidence of higher bond amounts and higher odds of pretrial detention for males relative to females (e.g., Demuth, 2003; Demuth & Steffensmeier, 2004; Goulette et al., 2015; Wooldredge, 2012; Wooldredge et al., 2015).

The bulk of our Level 1 measures consist of legally relevant case factors and criminal histories because these are consistently the strongest predictors of case dispositions and outcomes (Ulmer, 1997; Wooldredge et al., 2015). Legally relevant measures included most serious felony charge (F1 through F5, with F5 as the reference), total felony charges, specifications, and murder suspect (included because murder charges improved prediction beyond the most serious charge measures). Specifications included sentence enhancements for use of firearms or if the suspect was a repeat violent offender. Both specifications and murder suspect were excluded from the analyses of property and drug offenses only.

Two victim measures tapping the number of victims of violence and whether any victim injury was sustained during the offense were also included in the analyses of the pooled sample and the sample of violent crimes only. Values of zero were assigned to all cases not meeting the labeled criterion (i.e., no victims of violence and no victim injury). For the pooled sample, zeros were also assigned to victimless crimes. Finally, whether a defendant had served a prior prison sentence was included as an indicator of criminal history. This was the strongest predictor of both outcomes relative to other available measures including prior arrests, felony arrests, convictions, felony convictions, and the number of prior prison sentences. It was also superior to a factor reflecting all of these measures combined (Cronbach’s α = .72).

The measure of bond amount (log_e) was also included in the model of pretrial detention as a statistical control. Defendants who are unable to post bond also do not obtain pretrial release, and this factor may correlate with neighborhood effects.

The Level 2 measure of neighborhood disadvantage was measured at the block level for all available crime locations in the sample (N₂ = 820). Street blocks were chosen because they are smaller units than census tracts and therefore may better reflect a person’s definition of his or her “neighborhood” (see Taylor’s, 1997, argument applied to face blocks and how individuals are sometimes unaware of what goes on just one street over). Smaller geographic units tend to also be more homogeneous on residential population factors, thus reducing within-group variance while increasing between-group variance in population composition relative to larger units. This is also consistent with Hunter (1974), who argued that residential groups tend to define themselves in terms of relative differences from other groups. Moreover, the idea that persons have a stronger sense of neighborhood identification in groups of individuals with similar characteristics and interests (Hawley, 1950) implies focusing on more narrow boundaries.

Incident and resident addresses were geocoded using ArcGIS (ArcView 10) to identify the Federal Information Processing Standards (FIPS) codes for the available blocks in the sample. These codes were then used to merge U.S. census data on population composition. The primary drawback to relying on blocks instead of census tracts is that the more traditional SES indicators are not nearly as plentiful for blocks, so we were limited to constructing a factor from a handful of available measures. Specifically, a principal components analysis of a block’s percentages of African American residents, female-headed households, vacant residences, and renters produced a single factor (Cronbach’s α = .81). Although not a scale of SES per se, we examined the correlation between this factor and a more “traditional” factor created at the tract-level including the above four items in addition to the percentages of unemployed civilians, residents living at or below the poverty level, residents receiving public assistance, and adults without high school degrees. The correlation between the two factors was examined at the tract level (n = 387), revealing a correlation (r) of .74. In the absence of any economic indicators, it would be inappropriate to label this factor as “SES,” so we use the term “neighborhood disadvantage” based on the greater marginalization of African Americans relative to Whites and of one-parent relative to two-parent households, in addition to less social capital in areas with higher percentages of renters (who tend to be more transient than property owners) and abandoned structures (which might reflect weaker investment by city government in the economics of particular neighborhoods).

Due to our interest in examining race effects at the individual level, it is possible that including race in the Level 2 measure of disadvantage could have been a problem due to covariation between a defendant’s race and percent African American residents in a block. Therefore, we explored differences in findings for Level 2 measures with versus without percent African American for all of the models described below. Slight raw differences in the maximum likelihood (ML) coefficients emerged for both defendant’s race and neighborhood disadvantage in every model in conjunction with no differences in statistically significant effects at either p ≤ .01 or p ≤ .05. (Differences in the pooled models are noted in the next section; offense-specific differences available upon request from the first author.) Moreover, the Pearson correlation between the two disadvantage measures is r = .98. Based on the similarities in findings between the two measures, percent African American was kept in the factor due to a slightly higher Kaiser-Meyer-Olkin score (0.75 vs. 0.71), and because the available block indicators were few and constructing a factor with four versus three items seems slightly more robust. From a different perspective, it also allows us to speak more directly about two levels of possible racial bias (at both the individual and neighborhood levels), recognizing that any significant SES effects at Level 2 could be wrapped up in macro race effects.

Given the inclusion of race at both levels of analysis, it was important to explore other differences in our results such as whether neighborhood effects at Level 2 changed when race was omitted at Level 1, and whether race effects at Level 1 changed with neighborhood disadvantage omitted at Level 2. The Level 1 race effects were virtually identical without neighborhood disadvantage included, which is intuitive given the two-stage estimation procedure with each Level 1 model estimated first to generate the Level 2 dependent variable for estimation of the neighborhood-level effect. The Level 2 disadvantage effects were also very similar although not nearly identical as the Level 1 race effects (likely due to grand mean centering race at Level 1, as described below). As such, we display these neighborhood effects with and without Level 1 race effects included in the models.

To create the measure of status differential between a defendant’s neighborhood of residence versus the neighborhood in which the incident occurred, “disadvantage” was also measured for the street block of each defendant’s residence. The value for the incident block was subtracted from the value of a defendant’s residential block to create the difference scale. Higher values on the scale indicate that the crimes occurred in less disadvantaged neighborhoods relative to where the defendant lived at the time of the incident. Figure 1 displays the distribution of status differential for the sample. This was treated as a Level 1 measure for the analysis because it is defendant specific, based on the unique combination of where the defendant lived and the neighborhood of the incident.

Figure 1.

Distribution of status differences in disadvantage levels between incident street block versus suspect’s residential street block (N₁ = 2,677).

Statistical Analysis

Multilevel modeling was used due to the nested data and the potential for correlated error among persons living in the same neighborhoods. Defendants at Level 1 (N₁ = 2,677) were nested within blocks at Level 2 (N₂ = 820). Recent research has demonstrated the utility of nesting cases within judges (Johnson, 2006), prosecutors (Wooldredge et al., 2015), or even both (Kim, Spohn, & Hedberg, 2015), providing the ability to assess variance in a particular outcome attributable to each court actor as well as combinations of actors. However, our focus here does not involve parsing out the variance in pretrial dispositions between judges and prosecutors.

Generalized least-squares regression models were estimated for bond amount (log_e), with models estimated for the pooled sample as well as for defendants charged with property versus violent versus drug crimes. Bernoulli (binary logistic) regression models were estimated for pretrial detention. The software for the analysis was HLM 7.0 (Raudenbush, Bryk, & Congdon, 2011). Level 1 proportionate reduction in error (PRE) values cannot be computed for multilevel Bernoulli models based on the information provided in HLM, so only the overall model fit and Level 2 PRE statistics are displayed for these models (vs. the addition of the Level 1 PRE statistics for the bond amount models).

The first step in each analysis involved determining whether there was significant between-neighborhood variance in each outcome, reflected in the Level 1 intercepts. Significant differences in these intercepts at Level 2 allowed us to treat these intercepts as random rather than fixed, thereby permitting estimation of the effect of neighborhood disadvantage in each model. The second step of each analysis involved estimation of a Level 1 model with all defendant level measures grand mean-centered to allow compositional effects on the outcomes at the neighborhood level. The last step focused on the intercepts-as-outcome models, which included both Level 1 and Level 2 predictors. The difference in the Level 2 variance in each outcome “explained” at Steps 2 and 3 represents the amount of variance explained by neighborhood disadvantage beyond the compositional effects of the Level 1 predictors. Given the very similar findings for Level 1 effects at Steps 2 and 3, only the findings for the intercepts-as-outcome models are displayed here.

Results and Discussion

Table 2 displays the intercepts-as-outcome models of bond amounts for the pooled sample as well as for each subsample of offense groups. The unconditional models for all groups revealed significant variance in bond amounts at Level 2 (ranging from p < .001 for the pooled sample to p < .05 for drug offenses). This information revealed that there was significant Level 2 variance to be explained, thus permitting analyses of whether neighborhood disadvantage might account for some of this variance in each model.

Table 2.

Generalized Least Squares Models of Bond Amount (log_e).

Variables	Pooled	Property	Violent	Drugs
Variables	b (SE_b)	b (SE_b)	b (SE_b)	b (SE_b)
Level 1: Felony defendants
Male	0.40** (0.06)	0.40** (0.08)	0.64** (0.15)	0.35 (0.19)
Age 18-29	0.13** (0.08)	0.24 (0.13)	−0.03 (0.20)	0.33 (0.25)
Age 30-49	0.15 (0.08)	0.20 (0.13)	0.08 (0.20)	0.36 (0.25)
African American	0.10* (0.05)	0.08 (0.09)	0.36* (0.14)	0.04 (0.20)
Most serious charge
Felony 1	1.94** (0.09)	2.58** (0.12)	0.67 (0.54)	0.58 (0.32)
Felony 2	0.88** (0.09)	0.97** (0.10)	−0.28 (0.54)	0.40 (0.25)
Felony 3	0.67** (0.07)	0.72** (0.12)	−0.14 (0.57)	0.32 (0.17)
Felony 4	0.37** (0.07)	0.25* (0.10)	−0.39 (0.55)	0.01 (0.17)
Total charges (square root)	0.17** (0.05)	0.22** (0.08)	0.50** (0.14)	0.50** (0.19)
Murder suspect	1.61* (0.72)	—	1.56* (0.74)	—
Gun specification	0.54** (0.09)	—	0.61** (0.14)	—
Repeat violent offender specification	−0.06* (0.03)	—	0.49** (0.16)	—
Victim physically injured	0.64** (0.08)	—	0.45** (0.12)	—
No. of victims of violence	0.16** (0.05)	—	0.04 (0.06)	—
Prior prison term served	0.31** (0.05)	0.34** (0.08)	0.07 (0.11)	−0.07 (0.15)
Status difference	−0.06* (0.03)	−0.05 (0.05)	−0.08 (0.08)	−0.00 (0.10)
Level 1 variance in outcome	1.98**	1.84**	2.02**	1.41**
Level 1 proportion variance in outcome explained	0.45	0.46	0.34	0.06
N ₁	2,677	974	627	330
Level 2: Street blocks
Neighborhood disadvantage	−0.04 (0.03)	0.02 (0.05)	−0.02 (0.07)	−0.15 (0.11)
Constant	8.80	8.61	10.15	8.54
Neighborhood disadvantage without race at Level 1	−0.02 (0.03)	0.04 (0.04)	0.02 (0.06)	−0.14 (0.09)
Level 2 variance in outcome	0.22**	0.29**	0.15*	0.07*
Intraclass correlation (ρ)	0.10	0.14	0.07	0.05
Level 2 proportion variance explained	0.44	0.90	0.00	0.00
N ₂	820	525	393	232

p < .05. **p < .01.

For the pooled sample as well as for each offense group, neighborhood disadvantage was a nonsignificant predictor of bond amounts whether or not percent African American was included in the Level 2 measure (pooled sample b = −0.04 and −0.02, with and without percent African American included, respectively; SE_b = 0.03 for both estimates). By contrast, for the pooled sample only, the status differential between a defendant’s neighborhood and the incident neighborhood was statistically significant (p < .05). Contrary to our prediction, however, defendants accused of crimes in neighborhoods with greater disadvantage relative to their own were assigned higher bond amounts, controlling for the number and severity of criminal charges. Although the pooled effect of status differential was similar in magnitude for defendants accused of either property or violent crimes, the last two effects were nonsignificant due to the smaller samples and suggest that the magnitude of the overall effect was relatively weak in each of the three models. The effect was weakest and virtually null for defendants accused of drug offenses. The significant inverse effect in the pooled model, although relatively weak, still raises the possibility that judges might be looking at distance traveled to lower SES neighborhoods as somehow linked to offender motivation and the problem with feeding crime in more disadvantaged neighborhoods. Moreover, the significant effect of status differential in the pooled model only could actually reflect the influence of the 746 offenders in the full sample charged with two or more offense types, where court actors might consider these offenders to inflict the greatest damage to lower status neighborhoods (e.g., drug dealers who promote violence). By contrast, offenders in each subsample were charged with only one type of crime.

African American defendants were assigned significantly higher bond amounts relative to White defendants in the pooled sample, consistent with previous research on bond amounts in a different jurisdiction (Wooldredge, 2012). The magnitude of the race effect was weakest in the analysis of defendants charged with drug crimes, yet was strongest by far in the analysis of defendants charged with violent crimes. Even so, legal factors such as charge severity, prior imprisonment, and victim injuries were much stronger predictors of bond amounts in the pooled sample, underscoring previous observations of the greater relevance of legal factors for shaping pretrial decisions (Bock & Frazier, 1978; Goldkamp & Gottfredson, 1985; Gottfredson & Gottfredson, 1988; Nagel, 1983; Thomas, 1976; Walker, 2006). In light of this observation, and specific to the jurisdiction examined here, perhaps it is not surprising that neighborhood effects on bond amounts were not significant given that these amounts were so strongly influenced by offense severity, offense type, and criminal history (consistent with Walker’s, 2006, review of the bail literature and the general theme that legal factors primarily drive bail decisions). Discretion in bond amount decisions might be limited in this jurisdiction due to constraints imposed by bond schedules and following “usual amounts,” as described earlier. By contrast, neighborhood effects on pretrial detention (described next) were far more substantive and suggest that greater emphasis should be placed on the analysis of pretrial detention in terms of its contribution to discussions of extralegal disparities in pretrial dispositions.

The intercepts-as-outcome models of pretrial detention are displayed in Table 3. Consistent with bond amounts, there was significant variance in the odds of pretrial detention across the blocks examined (p < .05 for all four unconditional models).

Table 3.

Bernoulli Models of Pretrial Detention for the Pooled Sample and Offense Subsamples.

Variables	Pooled	Property	Violent	Drugs
Variables	e^b	e^b	e^b	e^b
Level 1: Felony defendants (N₁ = 2,677)
Bond amount log_e	2.39**	2.64**	2.61**	1.83**
Male	1.39*	1.83**	3.69**	0.64
Age 18-29	1.22	0.83	1.73	2.06
Age 30-49	1.18	0.90	1.00	2.42
African American	1.05	1.06	1.72*	0.93
Most serious charge
Felony 1	1.28	0.81	0.52	1.08
Felony 2	1.03	0.81	0.43	1.55
Felony 3	0.89	0.69	0.29	0.84
Felony 4	0.67*	0.62	0.56	0.79
Total charges (square root)	0.86	0.96	1.37	0.73
Murder suspect	2.04	—	3.27**	—
Gun specification	0.67*	—	0.95	—
Repeat violent offender specification	2.27**	—	3.12**	—
Victim physically injured	0.68*	—	0.95	—
No. of victims of violence	1.07	—	1.14	—
Prior prison term served	2.23**	3.08**	1.95**	2.39**
Status difference	1.22*	1.46**	1.20	1.25
N ₁	2,677	974	627	330
Level 2: Street blocks (N₂ = 820)
Neighborhood disadvantage	1.19*	1.30*	1.01	1.31
Constant	0.30	0.30	1.01	0.11
Neighborhood disadvantage without race at Level 1	1.14*	1.20*	1.01	1.07
Level 2 variance in outcome	0.18**	0.05**	0.03*	0.52*
Level 2 proportion variance explained	0.16	0.28	0.00	0.00
N ₂	820	525	393	232

p < .05. **p < .01.

In contrast to bond amounts, both neighborhood disadvantage and defendant’s status differential were significant predictors of pretrial detention and in the predicted directions for the pooled sample. The odds of detention were significantly higher for defendants charged with crimes committed in more disadvantaged neighborhoods (p < .05), and this finding held regardless of whether percent African American was included in the disadvantage index (e^b = 1.19 and 1.21, with and without percent African American included, respectively; p < .05 for both estimates). Odds of detention were also higher when the defendant’s neighborhood was more disadvantaged than the neighborhood of the incident (p < .05), even when controlling for the number and severity of charges. The significant finding for status differential (between the offender’s neighborhood and the incident neighborhood) is consistent with a “threat” perspective where judges may be more concerned with offenders who travel outside lower SES areas to commit crimes in higher status neighborhoods. By contrast, the significant finding for neighborhood disadvantage is consistent with a community justice perspective where court actors may also be concerned with “cleaning up” more crime-ridden areas by cracking down on offenders in general who commit their crimes in lower SES neighborhoods.

The finding for neighborhood disadvantage suggests that judges may be more apt to deny pretrial release and/or less inclined to assign ROR to crack down on offenders in more crime-ridden neighborhoods. Important to note is that an analysis of neighborhood disadvantage in the defendant’s own neighborhood revealed the same effect on pretrial detention, which is not surprising given that most defendants were accused of crimes either in their own neighborhoods or in areas with similar levels of disadvantage. Judges may perceive these defendants as greater threats to public safety in these areas, consistent with focal concerns theory.

The bond amount assigned to a defendant was the strongest predictor of pretrial detention for the pooled sample, where detention was more likely when bond was set higher. Even though neighborhood disadvantage was not a significant predictor of bond amounts, the latter still constitutes the primary barrier to pretrial release for defendants facing greater economic disadvantage. The findings for pretrial release therefore reflect two levels of disadvantage at both the defendant level (less able to post higher bonds) and at the neighborhood level (perhaps due to judges’ efforts to protect residents of more crime-ridden communities). Although a defendant’s race was not significantly linked to pretrial detention for the pooled sample, its significant effect on bond amounts implies that African Americans still face higher odds of pretrial detention by nature of facing higher bond amounts, on average (see Wooldredge et al., 2015, for a related discussion of possible indirect effects of a defendant’s race on pretrial detention).

The analysis of the pooled sample might mask important differences in these effects by offense type, so it is important to compare these findings with estimates from the offense-specific models. The model for property offenses revealed consistent findings of significant effects of a defendant’s status differential and incident neighborhood disadvantage on the odds of pretrial detention, both in predicted directions. The magnitude of each effect was also larger, with a status difference effect 2 times larger than for the pooled sample. The magnitude of each effect was also larger for drug offenses although neither reached statistical significance, perhaps due to the much smaller subsample. By contrast, it is clear that status differential and neighborhood disadvantage had no impact on the odds of pretrial detention for the pool of violent offenses. Yet, a defendant’s race was a significant predictor of pretrial detention only for the pool of violent offenses, where African Americans charged with these crimes were more likely to be detained. This raises the possibility that race somehow becomes tied to pretrial decisions involving violent offenses, whereas neighborhood factors are more relevant for nonviolent offenses. Property and drug offenses therefore drove the significant environmental effects for the pooled sample. Court actors may feel that they are better able to send a message to property and drug offenders given the greater premeditation generally associated with each type of crime.

In contrast to these differences by offense type, bond amount was one of the strongest predictors of pretrial detention regardless of offense type. These findings imply that bond amount constitutes the primary barrier to pretrial release for defendants facing greater economic disadvantage, regardless of their charged offenses. Important to note, however, is that African Americans charged with violence were significantly more likely to be detained prior to trial, consistent with expectations based on a racial threat perspective. Further investigation revealed that dropping bond amount from the models generated stronger race effects for the samples of defendants charged with property, violent, and any crimes. We attribute these differences to the significant correlation between race and bond amounts described above.

All told, our analysis suggests that neighborhood factors (both disadvantage and the difference in disadvantage “status” between the defendant’s neighborhood and the incident neighborhood) matter primarily for shaping pretrial detention rather than bond amounts, underscoring their relevance for further consideration in studies of pretrial detention. This difference might be attributable to how bond amounts are set in this jurisdiction (based in part on “usual amounts” for certain offenses, as described earlier), whereas detention is driven more by improving the odds of appearance in court and protection of the community. However, this is not to say that our previous speculation regarding the significant but counterintuitive inverse effect of status differential on bond amounts for the pooled sample should be dismissed. On the contrary, the possibility that court actors perceive defendants charged with multiple types of crimes as more dangerous to crime-ridden communities when they are motivated by their crimes to travel to those communities is worthy of consideration in future research.

Neighborhood disadvantage might be relevant to pretrial detention from a community justice perspective (Karp & Clear, 2000), in light of public pressures placed on the courts and prosecutors to clean up crime-ridden neighborhoods. Disadvantage might also be relevant from a focal concerns perspective (Demuth & Steffensmeier, 2004; Steffensmeier et al., 1998) given that our findings also held for a defendant’s neighborhood of residence. That is, the courts might perceive offenders from more disadvantaged neighborhoods as greater risks to public safety in those areas, and judges may be less inclined to assign bond or ROR in such cases given the criminal opportunities in these environments. Judges might prefer to err on the side of holding someone rather than return someone to a community where they are likely to offend again (and possibly appear in the media). This idea would be consistent with Vîlcică and Goldkamp’s (2015) finding that neighborhood SES was a significant predictor of pretrial crime in Philadelphia.

The significant positive effect of status differences between suspect and incident neighborhoods on pretrial detention also suggests support for a focal concerns perspective if judges are likely to perceive offenders who are willing to travel outside their neighborhoods to victimize residents of higher status neighborhoods as “more dangerous.” This is consistent with the idea of “more law” applied to offenders perceived as posing greater threats to populations supportive of the status quo (Black, 1976; Nobiling et al., 1998; Spitzer, 1975). Therefore, while judges may be more heavy-handed with offenders from more disadvantaged neighborhoods, they may kick this up a notch when those same offenders dare to commit crimes in more affluent areas of their city.

Implications

The findings for neighborhood effects on pretrial release suggest that pretrial disparities in the treatment of felony suspects may be compounded by social disadvantage at the community level. First, efforts to crack down on defendants who commit crimes in more disadvantaged neighborhoods are necessarily absorbing even more disadvantaged defendants into the criminal justice system because most of these offenders either live in these or similarly disadvantaged areas. Disparities in pretrial processing may therefore be twofold, operating at both the individual and aggregate levels. Second, the higher odds of detention for suspects who victimize residents of less disadvantaged areas relative to their own present yet another source of harsher treatment for these individuals. Whereas the neighborhood disadvantage effect is consistent with either a focal concerns (Steffensmeier et al., 1998) or community justice (Karp & Clear, 2000) perspective, the status differential effect is more consistent with Black’s (1976) behavior of law and the idea of how law is enforced more heavily when crime becomes more of a threat to residents of higher SES, also reflective of a “power threat” perspective (Blalock, 1967).

Our findings also have implications for discussions of how a defendant’s race might affect case dispositions. Consider the ways in which African Americans were disadvantaged in the jurisdiction under study: (a) African Americans faced higher bond amounts, and bond amount (in turn) was one of the stronger predictors of pretrial detention; (b) greater neighborhood disadvantage corresponded with higher odds of pretrial detention, and greater “disadvantage” here was defined in part by higher percentages of African American residents due to the level of segregation in this particular area; and (c) individuals from neighborhoods with higher percentages of African Americans faced higher odds of detention if they were accused of committing crimes in neighborhoods with lower percentages of African Americans.

The discussion to this point suggests that more severe treatment of offenders from lower SES neighborhoods is undesirable. However, from a community justice perspective, it is only fair to present the counterargument that noncriminal residents of lower SES neighborhoods might benefit from tighter formal controls over offenders who live and commit crimes in those neighborhoods. Also, court actors might feel justified to some extent in this approach if they perceive more social capital in higher SES neighborhoods where residents are better able to engage in self-help (e.g., through community action programs) and can rely less on police.

Given the oft cited link between pretrial detention and the odds of imprisonment for convicted defendants (Reitler et al., 2013), these findings also have implications for research on “cumulative disadvantage” in the courts (Ulmer, 2012). Recent studies offer evidence that greater disadvantages faced by some defendants in earlier decision points can generate even greater disadvantages at later decision points (Kutateladze et al., 2014; Starr & Rehavi, 2012; Schlesinger, 2007; Spohn, 2009; Stolzenberg, D’Alessio, & Eitle, 2013; Sutton, 2013; Wooldredge et al., 2015). For example, if lower SES defendants are more likely to be detained prior to trial, and pretrial detention also increases a convicted defendant’s odds of imprisonment, then the influence of SES on imprisonment is compounded through its effect on pretrial detention in conjunction with any unique effect of SES on sentencing decisions. In this way, the effects of neighborhood disadvantage and status differences on sentencing might be compounded through their impacts on pretrial detention in addition to any unique effects on sentencing (although this remains a question for future research). Wooldredge (2007) found higher odds of prison sentences for convicted felons residing in more economically depressed census tracts of seven urban jurisdictions. Although his study focused on defendants’ residential locations as opposed to crime locations, and on census tracts instead of street blocks, this observation suggests that it might be worthwhile to examine the effects of incident block-level disadvantage on sentencing as well as other stages of case processing.

Ecological factors are theoretically relevant to critical perspectives on criminal case processing because a defendant’s class status is shaped by both individual and community level characteristics. Considering that the vast majority of our sample committed crimes in areas of similar disadvantage to their own, our findings for pretrial detention also complement Rose and Clear’s (1998) commentary on how social networks in poor communities are disrupted by the high incarceration rates of residents in those communities. Social cohesion and the means of self-regulation among community residents are damaged with reductions in human capital.

The implications of our findings for policies to reduce disparities in pretrial detention based on community factors are not as clear-cut as the theoretical implications. Bail guidelines, for example, are designed to reduce individual-level disparities in bail amounts but are not designed to reduce community level variation in the treatment of criminal suspects. Strategies to reduce considerations of SES-related factors at the individual level may conflict with community programs to reduce crime in particularly dangerous areas of a city, given that the macro-level characteristics of these areas correlate with micro-level indicators of disadvantage. Specific to our study, the bilevel analysis of pretrial detention demonstrated that a defendant’s race does not have to coincide with higher or lower odds of detention while defendants from predominantly African American neighborhoods still face higher odds of pretrial detention.

Study Limitations and Directions for Future Research

Given the scarcity of related empirical studies moving beyond individual-level effects on pretrial dispositions (but see Vîlcică & Goldkamp, 2015; Williams, 2005), it is important to identify particular limitations of our analysis that should be addressed in future research. Perhaps the most important limitation and the most difficult to resolve involved the absence of measures capturing factors that might have led judges to set higher or lower bonds relative to “going rates” in particular cases and the influence of prosecutors’ recommendations, if any (e.g., demeanor of the accused, whether the suspect acted out at initial appearance, case evidence, etc.). Survey data would be the most likely source for constructing related measures, also capturing other potentially relevant factors such as court actors’ punitive ideologies and the structure of courtroom workgroups, assuming analyses of multiple jurisdictions. This last point is also a limitation of examining a single jurisdiction because such analyses cannot provide comparisons of counties that operate under bail guidelines as opposed to bail schedules or “going rates.” Another advantage to examining multiple jurisdictions also relates to the relevance of neighborhood disadvantage and the ability to see whether the effect of disadvantage varies significantly based on court factors (e.g., the effect might be weaker in jurisdictions with bail guidelines if these systems reduce judicial discretion).

Related analyses might also benefit from an ability to examine more specific offense groups. The limited sample size for our study only enabled analyses of general groups of violent versus drug versus property offenses. Separating drug crimes into selling versus possession only could be insightful, as might separating violent crimes involving firearms versus other weapons versus no weapons. More serious offenses might be associated with less discretion shown by judges in addition to much higher odds of denying bond altogether, potentially dampening the effects of neighborhood disadvantage on bond amounts and the odds of pretrial detention.

The ability to study different operational definitions of “neighborhoods” would also be useful, if the strength of neighborhood effects actually depends on how neighborhood boundaries are defined. Neighborhood disadvantage might have stronger effects on bond amounts when examining larger units than city blocks, if these larger units more accurately reflect how court actors themselves perceive more versus less crime-ridden areas. On the other hand, smaller units such as street segments might be more important for understanding bail decisions given that crime hot spots might be concentrated on only one street along two or three city blocks.

Exploring additional neighborhood factors that might account for between-neighborhood variance in bond amounts would also be important, given that our null findings beg the question of why these neighborhood differences were statistically significant. It could be that court actors are concerned with some but not all disadvantaged areas. Other equally disadvantaged neighborhoods might hold less interest because they are not as crime-ridden or there is less political pressure to “clean up” some of these areas. Indicators of “hot spots” might account for differences in bond amounts across city blocks that cannot be explained by SES factors per se.

Finally, and returning back to the presently unique contributions of these types of studies, the ability to move beyond an individual-level focus when studying multiple stages of case processing (not just bail and pretrial detention) would contribute greatly to the growing literature on cumulative disadvantage in the courts. Specific to our focus, these analyses will reveal whether harsher dispositions earlier in the system for defendants from poorer neighborhoods (e.g., pretrial detention) generate even more severe case outcomes (prison sentences), beyond any proximate effects of neighborhood status on sentencing that might occur due to citizen involvement in sentencing decisions. Examining both individual and aggregate effects on multiple stages of case processing would permit a much more comprehensive understanding of extralegal disparities in case processing at both the micro and macro levels in conjunction with how these factors operate directly and indirectly to shape the severity of treatment by the courts.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author(s) received financial support for this research through the County Prosecutors Office of the jurisdiction examined.

Author Biographies

John Wooldredge is a professor in the School of Criminal Justice at the University of Cincinnati. His research and publications focus on institutional corrections (crowding, inmate violence, inmate adaptation) and criminal case processing (sentencing and recidivism, extralegal disparities in case processing and outcomes).

James Frank is a professor at the University of Cincinnati. He received his JD from Ohio Northern University in 1977 and PhD from the School of Criminal Justice at Michigan State University in 1993. His primary research interests include understanding police behavior at the street level, the formation of citizen attitudes toward the police, and sentencing. He has published articles in Justice Quarterly, Police Quarterly, Journal of Criminal Justice, Criminal Justice Policy Review, Crime & Delinquency, and Policing: An International Journal of Police Strategy & Management.

Natalie Goulette is an assistant professor in the Department of Criminal Justice at the University of West Florida. She received her PhD from the School of Criminal Justice at the University of Cincinnati. Her research interests include extralegal disparities in case processing, the collateral consequences of criminal convictions, and the use of evidence-based practices in correctional programming.

References

Albonetti

(1987). Prosecutorial discretion: The effects of uncertainty. Law & Society Review, 21, 291-313.

Albonetti

(1989). Bail and judicial discretion in the District of Columbia. Sociology and Social Research, 74, 40-47.

Albonetti

(1991). An integration of theories to explain judicial discretion. Social Problems, 38, 247-266.

Ayres

Waldfogel

(1994). A market test for race discrimination in bail setting. Stanford Law Review, 46, 987-1047.

Black

(1976). The behavior of law. New York, NY: Academic Press.

Blalock

(1967). Toward a theory of minority-group relations. New York, NY: John Wiley.

Blumer

(1955). Reflections on theory of race relations. In Lind

(Ed.), Race relations in world perspective (pp. 3-21). Honolulu: University of Hawaii Press.

Bobo

Hutchings

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

Bock

E. W.

Frazier

(1978). Official standards versus actual criteria in bond dispositions. Journal of Criminal Justice, 5, 321-328.

10.

Chiricos

Bales

(1991). Unemployment and punishment: An empirical assessment. Criminology, 29, 701-724.

11.

Daly

(1989). Rethinking judicial paternalism: Gender, work-family relations, and sentencing. Gender & Society, 3, 9-36.

12.

Demuth

(2003). Racial and ethnic differences in pretrial release decisions and outcomes: A comparison of Hispanic, Black, and White felony arrestees. Criminology, 41, 873-907.

13.

Demuth

Steffensmeier

(2004). The impact of gender and race/ethnicity in the pretrial release process. Social Problems, 51, 222-242.

14.

Fagan

Davies

(2000). Street stops and broken windows: Race and disorder in New York City. Fordham Urban Law Journal, 28, 457-504.

15.

Foote

(1954). Compelling appearance in court: Administration of bail in Philadelphia. University of Pennsylvania Law Review, 102, 1031-1079.

16.

Fox

(2008). Applied regression and generalized linear models. Los Angeles, CA: Sage.

17.

Frazier

Bock

Henretta

(1980). Pretrial release and bail decisions: The effects of legal, community, and personal variables. Criminology, 18, 162-181.

18.

Goldkamp

(1979). Two classes of accused. Cambridge, MA: Ballinger Press.

19.

Goldkamp

Gottfredson

(1985). Policy guidelines for bail: An experiment in court reform. Philadelphia, PA: Temple University Press.

20.

Goldkamp

Gottfredson

Mitchell-Herzfeld

(1981). Bail decisionmaking: A study of policy guidelines. Albany: State University of New York Press.

21.

Gottfredson

(1988). Decisionmaking in criminal justice: Towards the rational exercise of discretion. New York, NY: Plenum Press.

22.

Goulette

Wooldredge

Frank

Travis

III. (2015). From initial appearance to sentencing: Do female defendants experience disparate treatment? Journal of Criminal Justice, 43, 406-417.

23.

Hawkins

(1981). Causal attribution and punishment for crime. Deviant Behavior, 2, 207-230.

24.

Hawley

(1950). Human ecology: A theory of urban structure. New York, NY: Ronald Press.

25.

Holmes

Daudistel

Farrell

(1987). Determinants of charge reductions and final dispositions in cases of burglary and robbery. Journal of Research in Crime & Delinquency, 24, 233-254.

26.

Holmes

Hosch

Daudistel

Perez

Graves

(1996). Ethnicity, legal resources, and felony dispositions in two southwestern jurisdictions. Justice Quarterly, 13, 11-30.

27.

Hunter

(1974). Symbolic communities: The persistence and change of Chicago’s local communities. Chicago, IL: University of Chicago Press.

28.

Johnson

(2006). The multilevel context of criminal sentencing: Integrating judge and county level influences in the study of courtroom decision making. Criminology, 44, 259-298.

29.

Karp

Clear

(2000). Community justice: A conceptual framework. In Friel

(Ed.), Criminal Justice 2000, Volume 2: Boundary changes in criminal justice organizations (pp. 323-368). Washington, DC: Office of Justice Programs.

30.

Katz

Spohn

(1995). The effect of race and gender on bail outcomes: Test of an interactive model. American Journal of Criminal Justice, 19, 161-184.

31.

Kim

Spohn

Hedberg

(2015). Federal sentencing as a complex collaborative process: Judges, prosecutors, judge-prosecutor dyads, and disparity in sentencing. Criminology, 53, 597-623.

32.

Klinger

(1997). Negotiating order in patrol work: An ecological theory of police response to deviance. Criminology, 35, 277-306.

33.

Kutateladze

Andiloro

Johnson

Spohn

(2014). Cumulative disadvantage: Examining racial and ethnic disparity in prosecution and sentencing. Criminology, 52, 514-551.

34.

LaFree

(1980). The effect of sexual stratification by race on official reactions to rape. American Sociological Review, 45, 842-854.

35.

LaFree

(1985). Official reactions to Hispanic defendants in the southwest. Journal of Research in Crime & Delinquency, 22, 213-237.

36.

Lizotte

(1978). Extralegal factors in Chicago’s criminal courts: Testing the conflict model of criminal justice. Social Problems, 25, 564-580.

37.

Maxwell

(1999). Examining the congruence between predictors of ROR and failures to appear. Journal of Criminal Justice, 27, 127-141.

38.

Nagel

(1983). The legal/extra-legal controversy: Judicial decisions in pretrial release. Law & Society Review, 17, 481-515.

39.

Nobiling

Spohn

DeLone

(1998). A tale of two counties: Unemployment and sentence severity. Justice Quarterly, 15, 459-485.

40.

Raudenbush

Bryk

Congdon

(2011). HLM7: Hierarchical linear and nonlinear modeling. Chicago, IL: Scientific Software International.

41.

Reiman

(2001). The rich get richer and the poor get prison. Boston, MA: Allyn & Bacon.

42.

Reitler

Sullivan

Frank

(2013). The effects of legal and extralegal factors on detention decision in U.S. district courts. Justice Quarterly, 30, 340-368.

43.

Richey-Mann

(1993). Unequal justice: A question of color. Bloomington: Indiana University Press.

44.

Rose

Clear

(1998). Incarceration, social capital, and crime: Implications for social disorganization theory. Criminology, 36, 441-479.

45.

Schlesinger

(2007). The cumulative effects of racial disparities in criminal processing. Journal of the Institute of Justice & International Studies, 7, 268-285.

46.

Smith

(1986). The neighborhood context of police behavior. In Reiss

Tonry

(Eds.), Crime and Justice, Volume 8: Communities and crime (pp. 313-341). Chicago, IL: University of Chicago Press.

47.

Smith

Visher

Davidson

(1984). Equity and discretionary justice: The influence of race on police arrest decisions. Journal of Criminal Law and Criminology, 75, 234-249.

48.

Spitzer

(1975). Toward a Marxian theory of deviance. Social Problems, 22, 638-651.

49.

Spohn

(2009). Race, sex, and pretrial detention in federal court: Indirect effects and cumulative disadvantage. Kansas Law Review, 57, 879-901.

50.

Spohn

Holleran

(2000). The imprisonment penalty paid by young, unemployed black and Hispanic male offenders. Criminology, 38, 281-306.

51.

Spohn

Spears

(1996). The effect of victim and offender characteristics on sexual assault case processing decisions. Justice Quarterly, 13, 649-679.

52.

Starr

Rehavi

(2012). Racial disparity in federal criminal charging andits sentencing consequences (Working Paper No. 12–002). Ann Arbor: University of Michigan Law School.

53.

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, 763-797.

54.

Stolzenberg

D’Alessio

Eitle

(2013). Race and cumulative discrimination in the prosecution of criminal defendants. Race and Justice, 3, 275-299.

55.

Stryker

Nagel

Hagan

(1983). Methodological issues in court research: Pretrial release decisions for federal defendants. Sociological Methods and Research, 11, 469-500.

56.

Sutton

(2013). Structural bias in the sentencing of felony defendants. Social Science Research, 42, 1207-1221.

57.

Taylor

(1997). Social order and disorder of street blocks and neighborhoods: Ecology, microecology, and the systemic model of social disorganization. Journal of Research in Crime & Delinquency, 34, 113-155.

58.

Terrill

Reisig

(2003). Neighborhood context and police use of force. Journal of Research in Crime & Delinquency, 40, 291-321.

59.

Thomas

(1976). Bail reform in America. Los Angeles: University of California Press.

60.

Ulmer

(1997). Social worlds of sentencing: Court communities under sentencing guidelines. Albany: State University of New York Press.

61.

Ulmer

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

62.

Vîlcică

Goldkamp

(2015). Bail prediction: Exploring the role of neighborhood context in Philadelphia. Criminal Justice and Behavior, 42, 1159-1182.

63.

Walker

(2006). Sense and nonsense about crime and drugs. Belmont, CA: Thomson Wadsworth.

64.

Williams

(2015). From bail to jail: The effect of jail capacity on bail decisions. American Journal of Criminal Justice, 41, 484-497.

65.

Wooldredge

(2007). Neighborhood effects on felony sentencing. Journal of Research in Crime & Delinquency, 44, 238-263.

66.

Wooldredge

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

67.

Wooldredge

Frank

Goulette

Travis

III. (2015). Is the impact of cumulative disadvantage on sentencing greater for black defendants? Criminology & Public Policy, 14, 187-223.