Sex-Based Disparities in Pretrial Release Decisions and Outcomes

Abstract

This study involved an examination of sex-based differences in judicial decision making during pretrial, an understudied stage of the criminal court process. We also examined whether defendants’ sex interacted with other legal and extralegal case characteristics, or the context of the county in which defendants were processed. Analyses of data collected from 62 large urban counties for the years 1994 through 2006 revealed that overall, women were treated more leniently than men throughout the pretrial process. Results also indicated that defendants’ sex interacted with other case characteristics to influence pretrial decisions and outcomes, and that judicial decisions were shaped by the sociopolitical context of the county in which these defendants were processed.

Keywords

pretrial sex-based disparities court community focal concerns

Empirical inquiry regarding the court processing of criminal defendants has historically focused on judicial decision making during the sentencing stage of criminal proceedings (Griffin & Wooldredge, 2006; Johnson, 2006; Spohn, 2000; Steffensmeier, Ulmer, & Kramer, 1998; Zatz, 2000). Pretrial, an earlier, less visible stage of the process has received much less attention (Demuth, 2003; Goldkamp, 1979). Yet, examination of judicial decision making during pretrial is important because these decisions affect defendants’ liberty, and incarceration can affect defendants’ job stability, family life, and the communities from which they are removed (Clear, 2007). Being held in pretrial detention has also been linked to harsher sanctions at subsequent stages of the criminal court process (Albonetti, 1991; Griffin & Wooldredge, 2006; Spohn, 2009), and so an understanding of the influences on judicial decision making during pretrial could shed light on whether these decisions might indirectly disadvantage particular groups of defendants at later stages of the criminal court process.

Judicial decisions made during pretrial are subject to less oversight than decisions made during other stages of the criminal court process (e.g., sentencing), which raises the possibility that these decisions will produce unwanted disparities. Considerable research has been directed at understanding the extent and sources of disparate treatment in the court processing of criminal defendants (e.g., Albonetti, 1991; Gottfredson & Gottfredson, 1988; Johnson, 2003; Mitchell, 2005; Spohn, 2000; Spohn & Holleran, 2000; Steen, Engen, & Gainey, 2005; Steffensmeier et al., 1998; Walker, 1993; Wooldredge, Griffin, & Rauschenberg, 2005; Zatz, 2000). Most of this research has centered on defendants’ race or ethnicity; less research has focused on sex-based disparities. Although researchers who have examined sex-based differences in sentencing outcomes have generally revealed that females are treated more leniently (e.g., Daly & Bordt, 1995; Griffin & Wooldredge, 2006; Koons-Witt, 2002), it is less clear whether sex-based disparities exist at the pretrial stage of the criminal court process (Demuth & Steffensmeier, 2004). Also unclear is whether judicial decision making during pretrial is influenced by the interaction of defendants’ sex with other individual or case characteristics, or the context of the county in which defendants are processed.

We address these shortcomings by examining judicial pretrial decisions for defendants processed in 62 urban counties across the United States. We examine (a) whether defendants receive differential treatment based on their sex, (b) whether defendants receive differential treatment based on the combination of their sex and other legal and extralegal characteristics, (c) whether treatment based on defendants’ sex differs across counties, and, if so, (d) whether these differences are shaped by the context of the county in which defendants are processed.

Theoretical Perspectives on Judicial Decision Making

Judicial decisions made during pretrial elicit the fundamental tension between judicial interests in controlling crime and providing due process equally under the law (Demuth, 2003; Walker, 1993). Judges consider defendants’ risk to the community and odds of appearing for subsequent proceedings, alongside their desire to minimize restrictions and/or penalties imposed on defendants who are legally innocent (Demuth, 2003; Goldkamp, 1979). Pretrial decisions are often reached with little or incomplete information, and these discretionary decisions are often made with less oversight than decisions made during later stages of the criminal court process (Demuth, 2003; Goldkamp, 1979). Still, enough similarities exist between these stages to inform expectations regarding the influences on judicial decision making during pretrial by drawing from the more extensive theoretical and empirical literature on decision making during other stages of the criminal court process.

Researchers of criminal court processes have generally found that legal criteria (e.g., offense type) account for most of the variation in judicial decisions; however, researchers have also acknowledged that judicial decision making is still a substantive process (e.g., Johnson, 2006; Johnson, Ulmer, & Kramer, 2008; Savelsberg, 1992; Ulmer, 2012). Albonetti (1986, 1987, 1991), for instance, has described how judges have an interest in controlling crime, but typically make decisions with limited information regarding defendants’ risk to reoffend. Inadequate information concerning defendants’ prospects for reform generates uncertainty during judicial decision making, which judges overcome by developing patterned responses to similar cases. Case outcomes, then, are influenced not only by formal considerations such as the type of offense or defendants’ criminal history but also by judges’ preconceptions of higher risk defendants. The basis for judges’ beliefs regarding high-risk defendants can be derived from their attributions regarding the causes of criminal behavior (Albonetti, 1991; Bridges & Steen, 1998; Hawkins, 1987). Defendants may be considered more culpable and a greater risk to reoffend if judges perceive the offenses they are alleged to have perpetrated were influenced more by personal factors as opposed to environmental factors. Judges may then be more likely to impose harsher outcomes (e.g., deny bail) on such defendants (Bridges & Steen, 1998).

Steffensmeier et al. (1998) have offered specific fields of reference that guide judges’ patterned responses or perceptual shorthand. Specifically, judges are directed by three focal concerns that include (a) their assessments of defendants’ blameworthiness, (b) their interests in protecting the community, and (c) the practical constraints related to individuals’ and organizational resources. These fields of reference may be particularly salient during pretrial because judges often make decisions with very little information regarding the facts surrounding cases. Blameworthiness refers to defendants’ culpability and the degree of harm caused to the victim. Defendants who are alleged to have committed more serious offenses, played a larger role in the commission of those offenses, or have more extensive criminal histories are typically considered more blameworthy and treated more harshly (Hartley, Maddan, & Spohn, 2007). During pretrial, defendants viewed as more blameworthy may receive harsher treatment (e.g., denied bail) because judges perceive them as a greater risk to the community and because these defendants have more to lose by reappearing (e.g., harsher sentences; Goldkamp, 1979).

Judicial interests in protecting the community evoke predictions regarding defendants’ risk of reoffending, which are shaped by stereotypes and symbolic markers associated with other individual and case characteristics. Because particular groups of individuals (e.g., males) are overrepresented within offender populations, judges may perceive defendants who possess these characteristics as more dangerous and with fewer prospects for reform. Thus, judges may treat defendants who are younger, male, minorities, or lower class more harshly.

Finally, judges are also constrained by the organizational realities of working in a functionally interdependent justice system, which compels them to consider local stakeholders who are affected by their decisions (e.g., jail capacity), as well as, external pressures from the larger community in which they are situated (e.g., political orientations). Judges also consider the practical and social costs of incarcerating particular defendants such as those who they perceive may have difficulty adapting to incarceration or defendants who have certain responsibilities (e.g., children). Defendants who are older, female, or White may receive leniency (e.g., lower bail) because these individuals are underrepresented within the offender population and may, therefore, be perceived as less able to cope with incarceration (Johnson, 2006; Steffensmeier et al., 1998; Wooldredge et al., 2005).

Sex-Based Disparities in Pretrial Decisions and Outcomes

The focal concerns perspective also predicts more lenient treatment of female defendants. Women generally commit less serious crimes and have shorter criminal histories than men (Steffensmeier, Kramer, & Streifel, 1993). Women are also underrepresented in offender populations (Steffensmeier & Allan, 1996), less likely to act alone when committing crimes, and frequently commit crimes with men (Alarid, Marquart, Burton, Cullen, & Cuvelier, 1996; Mullins & Wright, 2003; Steffensmeier & Terry, 1986). For these reasons, judges may treat female defendants more leniently because they consider females less blameworthy and/or lower risk to reoffend (Griffin & Wooldredge, 2006; Steffensmeier et al., 1998).

Female defendants could also receive leniency due to judicial perceptions regarding the social costs of sanctioning female defendants severely. Women are often viewed as caregivers and so imposing a large bail amount or incarcerating them could inhibit their ability to fulfill their familial responsibilities and breakup traditional family dynamics (Daly, 1987, 1989; Koons-Witt, 2002). Judges are also aware that many women who come to the attention of the justice system have been victimized (Browne, Miller, & Maguin, 1999; Harlow, 1999), and thus, may view female defendants as less able to adapt to incarceration (Steffensmeier et al., 1998). In support of these ideas, researchers have typically uncovered that female defendants are treated more leniently than male defendants during different stages of the criminal court process (e.g., Bushway & Piehl, 2001; Daly & Bordt, 1995; Demuth & Steffensmeier, 2004; Griffin & Wooldredge, 2006; Kruttschnitt & McCarthy, 1985; Spohn, 2009; Spohn & Holleran, 2000; Steffensmeier et al., 1993; Steffensmeier et al., 1998). Based on these perspectives and related findings we expect the following:

Hypothesis 1: Female defendants will receive more lenient treatment than male defendants throughout the pretrial process (e.g., lower bail amounts), net of other legal and extralegal factors.

Scholars have also hypothesized that although judges typically treat female defendants more leniently, not all females are treated equally (e.g., Demuth & Steffensmeier, 2004; Griffin & Wooldredge, 2006; Koons-Witt, 2002). Particular groups of women may be treated more harshly than other women; some may even be treated more harshly than men. For instance, women who are arrested for traditionally “male crimes” (e.g., robbery) or have lengthier criminal histories may be viewed as more blameworthy or a greater risk to the community, and so judges may treat them more harshly than other female defendants, and perhaps, even men (Nagel & Hagan, 1983; Steffensmeier et al., 1993). Commonly referred to as the “evil woman hypothesis,” this perspective suggests that women who do not conform to prescribed gender roles are especially blameworthy for their actions and are no longer deserving of the leniency typically extended toward women. Judges may punish these women in attempt to reinforce women’s traditional place in society (Chesney-Lind, 1989; Kruttschnitt, 1981). Thus, while the focal concerns perspective and evil woman hypothesis suggest that women who are arrested for more serious offenses or have more extensive criminal histories are treated more punitively than other women, the evil woman hypothesis posits that the effects of these characteristics may be more pronounced among women versus men. In contrast, the focal concerns perspective predicts no differences in the effects of these characteristics across sexes. To date, there has been limited support for the evil woman perspective (Daly & Tonry, 1997; Steffensmeier et al., 1993).

However, some research has suggested that extralegal characteristics (e.g., age, race) may condition judicial treatment of female defendants. Based on the focal concerns perspective, female defendants who are younger or older than the majority of defendants may be afforded further leniency because judges perceive that these women may be less able to cope with incarceration (Steffensmeier et al., 1998). Compared to White women, minority women may be treated more harshly because judges may perceive that minority women are better able to cope with the rigors of incarceration due to their overrepresentation within the female offender population (Demuth & Steffensmeier, 2004; Griffin & Wooldredge, 2006). Minority women are still underrepresented relative to minority men, however, so these women might receive more favorable treatment than comparable men.

In one of the few related studies, Demuth and Steffensmeier (2004) found that female defendants received more lenient treatment than men during each stage of the pretrial process. Compared to White female defendants, African American and Hispanic women were more likely to be detained prior to trial, and African American women were more likely to be held on bail. Regardless of their race/ethnicity, however, female defendants received more favorable treatment than men, although African American and Hispanic women were more likely to be held on bail than White women or men.

We build on Demuth and Steffensmeier’s (2004) findings in several important ways. First, we conduct similar analyses with more contemporary data. Second, we examine sex-specific samples to assess whether defendants’ sex interacts with other legal and extralegal variables besides race/ethnicity. Based on the focal concerns perspective and extant evidence discussed above, we expect the following:

Hypothesis 2: Female defendants who are arrested for more serious crimes, have lengthier criminal histories, are racial/ethnic minorities, or are between the ages of 21 and 29 will receive treatment that is less favorable than other female defendants.

Hypothesis 3: The relationship between defendants’ sex and pretrial outcomes will not be conditioned by other legal characteristics (e.g., being arrested for a violent offense will have a similar effect among women and men).

Hypothesis 4: The relationship between defendants’ sex and pretrial outcomes will be conditioned by other extralegal characteristics (e.g., being a racial/ethnic minority will have a stronger effect on more punitive pretrial outcomes among men vs. women).

Finally, we also move beyond Demuth and Steffensmeier’s (2004) by examining whether the treatment of female defendants varies across counties, and if so, whether this variation is shaped by the context of the county in which they were processed. Although researchers have considered the relevance of county context to explain sentencing decisions (i.e., incarceration), the effects of county contextual factors on pretrial decisions and outcomes have not been examined.

Main and Moderating Effects of County Context on Pretrial Decisions and Outcomes

Researchers of criminal courts have uncovered between-county variation in case processing, and these scholars have described county courts as communities comprising interdependent participants and stakeholders who hold common beliefs regarding, for example, how cases should be processed (e.g., Eisenstein, Flemming, & Nardulli, 1988; Ulmer & Johnson, 2004). Within these court communities, a unique legal and organizational culture exists that shapes an informal consensus regarding “going rates” or patterned responses that are applied to defendants with similar case characteristics. Variation in court outcomes can be shaped by differences in court organizations, influence of key stakeholders, as well as the broader sociopolitical environments in which courts are situated (Eisenstein et al., 1988; Helms & Jacobs, 2002; Ulmer, Bader, & Gault, 2008; Ulmer & Johnson, 2004). Given the similarities in court organizations under study here (e.g., large, urban), we focus primarily on characteristics of the sociopolitical environments of these courts.

The focal concerns perspective acknowledges that judicial decision making may be affected by external community pressures which vary across counties (e.g., Ulmer & Bradley, 2006; Ulmer & Johnson, 2004). We consider potential sources of between-county variation in pretrial decisions and outcomes that might affect judges’ perceptions of defendants’ blameworthiness, interests in protecting the community, and concerns about organizational resources and the social costs of treating certain defendants harshly. We assess how county context shapes judicial decisions during pretrial, and in particular, how county contextual factors may shape judicial treatment of female defendants relative to male defendants across these counties. We expect that,

Hypothesis 5: The relationship between defendants’ sex and pretrial outcomes will vary across counties; this relationship will be conditioned by characteristics of the county in which defendants are processed.

Based on the focal concerns perspective, judges have an interest in protecting the community, and the degree to which serious crime is a problem may influence their perspectives regarding sanctioning (Britt, 2000; Myers & Talarico, 1987). For instance, higher rates of violent crime within communities could indicate that institutions of social control have broken down. Judges may respond by denying bail more often or imposing higher bail amounts in an attempt to control and/or reduce crime (Fearn, 2005). Similarly, in counties with higher rates of female crime, judges may be less likely to afford female defendants leniency. Thus,

Hypothesis 5a: Female defendants processed in counties with higher rates of female violent crime will receive harsher treatment during pretrial than female defendants processed in counties with lower rates of female violent crime.

Ecological areas that contain higher rates of broken homes, persons living in poverty, and female-headed households with children are structurally disadvantaged (Sampson & Groves, 1989; Sampson & Laub, 1993; Sampson, Raudenbush, & Earls, 1997), and might evoke judicial perceptions regarding a breakdown in informal social controls (e.g., Britt, 2000; Sampson & Laub, 1993). Given the judges’ focal concern regarding community protection, judges who process cases in structurally disadvantaged areas might respond to these environmental cues by treating most defendants who come to their attention more severely (Britt, 2000; Sampson & Laub, 1993). Also consistent with the focal concerns perspective, however, judges might offer more lenient treatment for women processed in structurally disadvantaged counties. In counties characterized by structural or “household” disadvantage, judges may view women as less blameworthy because they attribute blame for their offending to environmental conditions, such as economic marginalization or familial instability. Judges may also view women as able to provide informal control over children, and consequently consider the related social costs associated with incarcerating women (Daly, 1987, 1989; Koons-Witt, 2002). Therefore, we predict:

Hypothesis 5b: Female defendants processed in counties with higher levels of household disadvantage will receive more lenient treatment during pretrial compared with female defendants processed in counties with lower levels of household disadvantage.

The focal concerns perspective holds that court communities are sensitive to local environmental pressures, including political and religious orientations (Fearn, 2005; Ulmer & Bradley, 2006; Ulmer et al., 2008). In some jurisdictions, judges are elected officials who are cognizant of and respond to the community’s predominant sentiment surrounding criminal punishment (Fearn, 2005; Ulmer & Johnson, 2004). Judges are also members of their community, and may be influenced by pervasive political and religious ideologies about crime and justice issues (Fearn, 2005). Areas with higher densities of residents who hold conservative values and beliefs (political or religious) often lean more toward law-and-order and tend to be more punitive (Applegate, Cullen, Fisher, & Vander Ven, 2000; Helms & Jacobs, 2002; Payne, Gainey, Triplett, & Danner, 2004; Unnever & Cullen, 2006; Unnever, Cullen, & Applegate, 2005), and so judges situated in more conservative counties may treat all defendants, including females, who come to their attention more harshly. Based on the focal concerns perspective, however, female defendants would still be considered less blameworthy than men because judges working in conservative counties may be more likely to endorse traditional paternalistic views regarding gender-specific roles for women (Helms & Jacobs, 2002). To date, research is mixed regarding the effects of conservative ideology on judicial sentencing practices (Fearn, 2005; Helms & Jacobs, 2002; Johnson, 2006; Ulmer, 2012; Ulmer et al., 2008; Ulmer & Johnson, 2004). No studies have examined the conditioning effects of conservative ideology on the relationship between defendants’ sex and pretrial outcomes.

Hypothesis 5c: Female defendants processed in counties where conservative ideology is more pervasive will receive harsher treatment during pretrial than female defendants processed in counties where conservative ideology is less prevalent.

Finally, the focal concerns perspective suggests that judges are constrained by their sensitivity to the demands their actions place on other organizations within their court community (Eisenstein et al., 1988; Johnson, 2006; Ulmer & Johnson, 2004). For instance, judges may deny bail less often, grant release more frequently, or impose lower bail amounts if the jail space in their county is scarce. This is because judges’ views regarding defendants’ blameworthiness or risk to the community can be affected by the constraint of having limited jail space at their disposal (Steffensmeier et al., 1998; Ulmer & Johnson, 2004). Because judges may view women as less blameworthy in counties with limited jail space, we hypothesize:

Hypothesis 5d: Female defendants processed in counties with jails that have less available bed space will receive more lenient treatment during pretrial than female defendants processed in counties with jails that have more available bed space.

Method

This study involves an examination of whether judicial decision making during pretrial differs for male and female defendants within and across urban counties. The target population for the study included all of the defendants processed in the nation’s 75 largest counties.

Sample

The defendant data used for the study were compiled from the State Court Processing Statistics (SCPS) for the years 1994-2006. The SCPS are collected for the Bureau of Justice Statistics biennially by the U.S. Census Bureau and provide detailed information on felony cases filed in 40 of the nation’s 75 largest counties. The SCPS involves a two-stage sampling design (counties followed by defendants) that has been described in detail elsewhere (e.g., Cohen & Kyckelhahn, 2010). Sample weights were derived by the Bureau of Justice Statistics that reflected the inverse of each defendant’s odds of selection. The weights were normalized (N_weighted = N_unweighted) and applied to the analyses reported here.

The combined sample size for the SCPS collected for the years 1994-2006 was 108,067 felony defendants processed in 67 counties. Due to the focus on judicial decision making during pretrial, we removed cases that were closed before pretrial or did not contain information regarding the pretrial stage of the process (N = 4,407). Next, we evaluated the remaining cases for missing data. The focus on defendants and counties required evaluation of the samples collected within each county for each year the SCPS data were collected. We observed that some counties did not report all of the required information (e.g., prior failure to appear) for particular years. For those years, these counties and the cases within them were removed. The same counties and related cases were retained for other years, however, so long as they reported complete information. For instance, if a county was included in the SCPS in both 2000 and 2006, but did not provide information regarding defendants’ prior history of failing to appear in year 2000, all of the cases for the county in year 2000 were removed, while the cases provided by the county in year 2006 were retained. Only the counties that provided data that were determined to be representative of their within-county target populations for the particular year of data collection were retained. Once we were confident that the remaining counties had adequate representation for generalization, we conducted a listwise deletion of the remaining cases in the combined data set that were missing information on relevant variables. This two-stage process reduced the sample to 77,688 felony defendants processed in 62 counties. There were no significant differences (p ≤ .01) between the final sample and the full sample (without missing data removed) for the measures age, sex, race, offense type, prior felony convictions, and all of the outcome measures.¹ The final samples (pooled, male, and female) are described in Table 1.

Table 1.

Sample Means and Standard Deviations (Unweighted) for Samples of Felony Defendants.

	Pooled	Female	Male
	M (SD)	M (SD)	M (SD)
Outcomes
Denied bail	0.06 (0.25)	0.04 (0.20)	0.07 (0.26)
ROR	0.17 (0.38)	0.24 (0.43)	0.16 (0.37)
Bail amount (logged)	9.14 (1.83)	8.70 (1.74)	9.22 (1.83)
Held on bail	0.50 (0.50)	0.38 (0.48)	0.52 (0.50)
Legal variables
Murder	0.01 (0.09)	0.00 (0.06)	0.01 (0.09)
Robbery	0.06 (0.23)	0.03 (0.17)	0.06 (0.25)
Assault	0.12 (0.32)	0.12 (0.32)	0.12 (0.32)
Other violent offense	0.06 (0.23)	0.04 (0.19)	0.06 (0.24)
Drug trafficking	0.17 (0.37)	0.13 (0.34)	0.17 (0.38)
Other drug offense	0.20 (0.40)	0.23 (0.42)	0.19 (0.40)
Public order offense	0.09 (0.29)	0.06 (0.23)	0.10 (0.30)
Property offense (ref.)	0.30 (0.46)	0.39 (0.49)	0.28 (0.45)
Multiple offenses	0.53 (0.50)	0.51 (0.50)	0.54 (0.50)
Prior felony conviction	0.41 (0.49)	0.29 (0.45)	0.43 (0.50)
Prior failure to appear	0.32 (0.47)	0.27 (0.44)	0.34 (0.47)
Active criminal justice status	0.34 (0.47)	0.27 (0.45)	0.36 (0.48)
Extralegal variables
Female	0.18 (0.38)	—	—
African American	0.45 (0.50)	0.46 (0.50)	0.45 (0.50)
Hispanic	0.23 (0.42)	0.15 (0.36)	0.24 (0.43)
Other race/ethnicity	0.02 (0.15)	0.02 (0.14)	0.02 (0.15)
White (ref.)	0.30 (0.46)	0.37 (0.48)	0.28 (0.45)
Age ≤ 20	0.18 (0.39)	0.12 (0.33)	0.20 (0.40)
Age 21-29 (ref.)	0.34 (0.47)	0.33 (0.47)	0.34 (0.47)
Age 30-49	0.43 (0.50)	0.50 (0.50)	0.42 (0.49)
Age ≥ 50	0.05 (0.22)	0.04 (0.20)	0.05 (0.22)
Year 1994	0.11 (0.32)	0.10 (0.30)	0.12 (0.32)
Year 1996	0.13 (0.33)	0.12 (0.33)	0.13 (0.33)
Year 1998	0.16 (0.36)	0.17 (0.37)	0.16 (0.36)
Year 2000 (ref.)	0.16 (0.37)	0.17 (0.37)	0.16 (0.36)
Year 2002	0.14 (0.35)	0.15 (0.35)	0.14 (0.35)
Year 2004	0.16 (0.36)	0.16 (0.37)	0.16 (0.36)
Year 2006	0.14 (0.35)	0.14 (0.35)	0.14 (0.35)
N ₁	77,688	13,811	63,877
County-level variables
Female violent crime rate	72.53 (60.50)
Household disadvantage^a	0.00 (0.99)
% female-headed households with children	0.12 (0.04)
% female-headed households in poverty	0.05 (0.03)
% females separated or divorced	0.19 (0.03)
Conservative ideology^a	0.00 (0.90)
% Republican	0.40 (0.12)
% Evangelical	0.09 (0.07)
Jail crowding	1.01 (0.23)
N ₂	62

Note. ROR = release on recognizance.

Scale created via factor analysis.

Measures

The outcome measures included judges’ decisions regarding bail, release on recognizance (ROR), bail amount, and whether defendants were held on bail. The decisions to deny bail and grant ROR were operationalized as dichotomous variables. Judges decisions to grant ROR were only considered for those defendants not denied bail (n = 72,645). The bail amount decision and held on bail outcome were only assessed for those defendants who were not denied bail, granted ROR, or given a nonfinancial release (n = 47,923). Bail amount was measured as a continuous variable reflecting the dollar amount judges set for bail. The distribution of bail amount was skewed so the natural log of the scale was used. Finally, held on bail, was measured as a dichotomous variable. Although held on bail is not a judicial decision, our interest was to determine whether the bail amount set by judges influenced whether defendants were detained.²

The legal variables used in the analyses included dichotomous measures reflecting each defendant’s most serious arresting offense: murder, robbery, assault, other violent offense, drug trafficking, other drug offense, and public order offense.³ Property offenses were treated as the reference category for the analyses. We also included a dichotomous measure of whether a defendant was arrested for multiple offenses. A measure of the number of offenses each defendant was charged with was available in the data set; however, bivariate analyses revealed that the dichotomous measure was a stronger predictor of the outcomes in this study. Defendants’ criminal history was measured with three dichotomous variables: prior felony conviction, prior failure to appear, and active criminal justice status. Prior felony conviction was included instead of other measures of prior record available in the data set (e.g., prior incarceration) because it maintained the strongest zero-order relationship with the majority of the outcomes examined in the study. Active criminal justice status measured whether the defendants were currently on probation or parole, or had been released on bail in another case. The extralegal variables included in the analyses were dichotomous measures of each defendant’s sex (female), race/ethnicity (African American, Hispanic, and Other race/ethnicity), and age (≤20, 30-49, and >50). White and the age group 21 to 29 were used as the reference categories for the analyses. Finally, dichotomous measures reflecting the year each defendant was processed were included as statistical controls. The year 2000 was treated as the reference category.

We also included four county-level variables. Female violent crime rate was created using sex-specific arrest data retrieved from the 2000 Uniform Crime Report. Household disadvantage was a scale that included county-level indicators of the percentage of residents living in female-headed households with children, the percentage of families that were female-headed and living in poverty, and the percentage of females who were separated or divorced (α = .95). These items were obtained from the 2000 Census. Conservatism was a scale that included the percentage of residents who voted for the Republican candidate in the 2000 election and the percentage of residents adhering to Evangelical beliefs (Pearson’s r = .50, α = .62). The percentage of residents who voted Republican was obtained from the Atlas of U.S. Presidential Elections. The percentage of residents adhering to Evangelical beliefs was obtained from the 2000 Religious Congregations and Membership Study. The household disadvantage and conservatism scales were created via a factor analysis that involved maximum likelihood extraction and varimax rotation.⁴ Jail crowding (average daily population/rated capacity) was obtained from the 1999 National Jail Census.⁵ For counties that had a combined jail and prison system, the combined jail and prison population and corresponding capacity was used.⁶

Finally, examination of several of the outcomes for this study involved the analysis of samples that were conditional upon decisions made at earlier stages in the pretrial process (e.g., ROR decision). To adjust for potential sample selection effects (e.g., Bushway, Johnson, & Slocum, 2007; Demuth, 2003), a correction factor was included in each of these models. For the analyses of the ROR decision and held on bail outcome, we applied a modified version of the Heckman two-step procedure that has been adapted for nonlinear outcomes (Dubin & Rivers, 1990; Greene, 2005). A correction factor generated via the Heckman two-step procedure was included in the analysis of the bail amount decision (Bushway et al., 2007; Heckman, 1976).⁷

Statistical Analysis

The hierarchical structure of the data (defendants nested within counties) required the use of multilevel modeling techniques. Specifically, these techniques permitted us to correct for the correlated error among defendants nested within counties and base hypothesis tests on the appropriate sample sizes (defendants vs. counties). Hierarchical Bernoulli regression was used to model the decisions to deny a defendant bail, grant a defendant ROR, and the held on bail outcome. Hierarchical linear regression was used for the analysis of the logged bail amount.

The analyses proceeded in several stages, which were executed using the software package HLM 7.0 (Raudenbush, Bryk, Cheong, & Congdon, 2011). First, unconditional models revealed whether there were significant differences in pretrial decisions and outcomes across counties (p ≤ .05). Next, random coefficient models (with only the Level 1 variables) were estimated to assess whether the relationships between the defendant-level variables and the outcomes varied across counties. Initially, all of the Level 1 relationships were allowed to vary across counties; however, those relationships that did not vary significantly (p > .05) were ultimately treated as fixed. With the exception of the year variables, all Level 1 variables were group mean centered to restrict the defendant-level models to explain within-county differences only. While this technique may result in more liberal estimates of the Level 2 main effects, it also reduces the odds of finding spurious Level 1 effects that may be due to unmeasured differences that may also be related to compositional differences across counties. Group mean centering also yields Level 1 estimates that are restricted to within-aggregate variation, which is preferable when assessing cross-level effects, such as in the current study. In contrast, the year variables were grand mean centered to control for unmeasured temporal influences at Level 2. The county-level variables were then added to the models to assess their effects on the Level 1 intercepts. Finally, the county-level effects on the Level 1 slopes depicting the relationship between defendants’ sex and each pretrial outcome were estimated. These analyses revealed whether the relationships between sex and pretrial decisions and outcomes were moderated by county context.

Separate defendant-level analyses were also conducted with only the male and female samples. The coefficients derived from these sex-specific analyses were compared using the equality of coefficients test developed by Clogg, Petkova, and Haritou (1995). The results of these z tests revealed whether the magnitude of effects generated for the predictor variables were greater for women versus men. Brame, Paternoster, Mazerolle, and Piquero (1998) demonstrated how this particular test can also be applied to compare maximum likelihood coefficients.

Results and Discussion

The results of the defendant-level analyses of the pooled sample are presented in Table 2. Tables 3 and 4 contain the analyses of the sex-specific samples and the results of the corresponding equality of coefficients tests, denoted by the z scores. Only the z scores that resulted from tests indicating a significant difference in the magnitude of the effects are reported. The county-level main effects and the county-level effects on the Level 1 relationship between defendants’ sex and each pretrial outcome are displayed in Table 5.

Table 2.

Random Coefficient Models of Pretrial Outcomes for the Pooled Sample (Standard Errors in Parentheses).

	Denied bail	ROR	Bail amount (logged)	Held on bail
Intercept	−1.79	−1.23	8.87	−0.15
Legal variables
Murder	1.78 ***	−0.74***	2.72 ***	−0.80***
	(0.17)	(0.12)	(0.19)	(0.22)
Robbery	0.21 ***	−0.54***	1.06 ***	−0.07
	(0.05)	(0.08)	(0.08)	(0.06)
Assault	0.08 **	−0.41***	0.56 ***	−0.60***
	(0.03)	(0.06)	(0.08)	(0.06)
Other violent offense	0.09 ***	−0.39***	0.83 ***	−0.67***
Other violent offense	(0.03)	(0.06)	(0.08)	(0.08)
Drug trafficking	−0.02	−0.21***	0.55 ***	−0.47***
Drug trafficking	(0.01)	(0.04)	(0.06)	(0.05)
Other drug offense	−0.06***	0.14	−0.30***	−0.28***
Other drug offense	(0.01)	(0.06)	(0.06)	(0.05)
Public order offense	−0.06**	−0.12***	−0.03	−0.51***
Public order offense	(0.02)	(0.04)	(0.05)	(0.05)
Multiple offense	−0.004	−0.03	0.22 ***	0.10 **
Multiple offense	(0.02)	(0.04)	(0.05)	(0.03)
Prior felony conviction	0.15 ***	−0.24***	0.24 ***	0.32 ***
Prior felony conviction	(0.02)	(0.04)	(0.05)	(0.03)
Prior failure to appear	0.02	−0.04	0.04	0.20 ***
Prior failure to appear	(0.02)	(0.02)	(0.04)	(0.04)
Active criminal justice status	0.38 ***	−0.25***	0.05	0.22 ***
Active criminal justice status	(0.05)	(0.07)	(0.05)	(0.06)
Bail amount (logged)	—	—	—	0.39 ***
Bail amount (logged)	—	—	—	(0.02)
Extralegal variables
Female	−0.07***	0.26 ***	−0.29***	−0.08
	(0.01)	(0.04)	(0.07)	(0.03)
African American	0.01	0.01	0.06	0.28 ***
African American	(0.01)	(0.03)	(0.03)	(0.03)
Hispanic	0.07 ***	−0.06	0.13 **	0.19 ***
	(0.02)	(0.03)	(0.04)	(0.04)
Other race/ethnicity	0.04	−0.07	0.06	−0.12
	(0.02)	(0.06)	(0.07)	(0.07)
Age ≤ 20	−0.05***	0.03	−0.0003	0.12 ***
	(0.01)	(0.02)	(0.03)	(0.03)
Age 30-49	−0.004	0.02	−0.01	0.11 ***
	(0.01)	(0.01)	(0.03)	(0.03)
Age ≥ 50	−0.004	0.14***	−0.18	−0.15**
	(0.01)	(0.03)	(0.08)	(0.06)
Inverse Mills ratio	—	0.72	−0.51	−0.92***
Inverse Mills ratio	—	(0.35)	(0.23)	(0.20)
χ²	4,192.89	12,002.15	22,860.69	4,166.94
N ₁	77,688	72,645	47,923	47,923

Note. ROR = release on recognizance. Models also include six dummy variables for the year defendants were processed. Hierarchical Bernoulli models predicting the decision to deny a defendant bail, decision to release a defendant on his or her own recognizance, and held on bail. Hierarchical linear model predicting bail amount. Italicized coefficients indicate that the effect varies across counties.

p ≤ .01. ***p ≤ .001.

Table 3.

Random Coefficient Bernoulli Models of Denied Bail and ROR Decisions for the Sex-Specific Samples (Standard Errors in Parentheses).

	Denied bail			ROR
	Female	Male	z test	Female	Male	z test
Intercept	−2.99	−1.76		−1.01	−1.30
Legal variables
Murder	2.40 ***	1.84 ***		−3.15***	−0.64***	−9.35
	(0.36)	(0.17)		(0.24)	(0.12)
Robbery	0.87***(0.14)	0.19 *** (0.05)	4.57	−.98 ***(.14)	−0.47***(0.08)	−3.16
Assault	0.22***	0.09 **		−0.57***	−0.34***
	(0.06)	(0.03)		(0.13)	(0.06)
Other violent offense	0.15	0.10 **		−0.31***	−0.35***
Other violent offense	(0.12)	(0.03)		(0.09)	(0.06)
Drug trafficking	0.19**	−0.03	3.02	−0.45***	−0.17***	−2.84
Drug trafficking	(0.07)	(0.02)		(0.09)	(0.04)
Other drug offense	0.06	−0.07***	3.61	0.01	0.18 **
Other drug offense	(0.03)	(0.02)		(0.06)	(0.06)
Public order offense	−0.08	−0.03		0.03	−0.09**
Public order offense	(0.16)	(0.02)		(0.11)	(0.04)
Multiple offenses	−0.04	0.005		−0.01	−0.05
Multiple offenses	(0.04)	(0.02)		(0.06)	(0.03)
Prior felony conviction	0.49***	0.15 ***	5.38	−0.38***	−0.23***
Prior felony conviction	(0.06)	(0.02)		(0.09)	(0.04)
Prior failure to appear	0.21***	0.01	4.47	−0.13**	−0.03
Prior failure to appear	(0.04)	(0.02)		(0.05)	(0.03)
Active criminal justice status	1.15 ***	0.38 ***	9.86	−0.35***	−0.20**
Active criminal justice status	(0.06)	(0.05)		(0.09)	(0.08)
Extralegal variables
African American	−0.11***	0.02	−4.11	0.08	−0.01
African American	(0.03)	(0.01)		(0.05)	(0.03)
Hispanic	0.05	0.08 ***		−0.07	−0.08
	(0.04)	(0.02)		(0.04)	(0.04)
Other race/ethnicity	−0.02	0.04		−0.12	−0.07
Other race/ethnicity	(0.13)	(0.03)		(0.17)	(0.05)
Age ≤ 20	−0.26***	−0.05**	−4.70	−0.01	0.05
	(0.04)	(0.02)		(0.06)	(0.02)
Age 30-49	−0.03	−0.002		−0.004	0.05 **
	(0.03)	(0.01)		(0.03)	(0.02)
Age ≥ 50	−0.05	−0.002		0.17	0.15***
	(0.10)	(0.02)		(0.07)	(0.03)
Inverse Mills ratio	—	—		0.06	0.47
Inverse Mills ratio	—	—		(0.49)	(0.38)
χ²	237.26	3,905.50		3,127.18	9,537.84
N ₁	13,811	63,877		13,255	59,390

Note. ROR = release on recognizance. Models also include six dummy variables for the years defendants were processed. Italicized coefficients indicate that the effect varies across counties.

p ≤ .01. ***p ≤ .001.

Table 4.

Random Coefficient Models of Bail Amount and Held on Bail for the Sex-Specific Samples (Standard Errors in Parentheses).

	Bail amount (logged)			Held on bail
	Female	Male	z test	Female	Male	z test
Intercept	8.41	8.95		−0.61	−0.08
Legal variables
Murder	3.14 ***	2.60 ***		0.04	−0.89***
	(0.49)	(0.19)		(0.38)	(0.20)
Robbery	1.04***	1.05 ***		0.31**	−0.19**	3.73
	(.13)	(0.08)		(0.12)	(0.06)
Assault	0.39 **	0.56 ***		−0.39***	−0.68***	2.73
	(0.13)	(0.07)		(0.08)	(0.07)
Other violent offense	0.26	0.87 ***	−3.78	−0.48***	−0.78***
Other violent offense	(0.14)	(0.08)		(0.14)	(0.08)
Drug trafficking	0.50 ***	0.53 ***		−0.11	−0.53***	4.45
Drug trafficking	(0.13)	(0.06)		(0.08)	(0.05)
Other drug offense	−0.17**	−0.33***		0.08	−0.33***	3.50
Other drug offense	(0.06)	(0.08)		(0.11)	(0.04)
Public order offense	0.08	−0.07		0.06	−0.60***	4.74
Public order offense	(0.06)	(0.04)		(0.13)	(0.05)
Multiple offenses	0.11	0.23 ***		0.14	0.09**
Multiple offenses	(0.06)	(0.05)		(0.07)	(0.03)
Prior felony conviction	0.15	0.25 ***		0.52***	0.28***	2.68
Prior felony conviction	(0.08)	(0.05)		(0.08)	(0.04)
Prior failure to appear	0.25	0.02		0.43***	0.18***	3.47
Prior failure to appear	(0.10)	(0.04)		(0.06)	(0.04)
Active criminal justice status	−0.002	0.06		0.18	0.21 ***
Active criminal justice status	(0.05)	(0.05)		(0.08)	(0.05)
Bail amount (logged)	—	—		0.39 ***	0.40***
Bail amount (logged)	—	—		(0.03)	(0.02)
Extralegal variables
African American	−0.06	0.08		0.25***	0.30***
African American	(0.06)	(0.04)		(0.07)	(0.03)
Hispanic	−0.03	0.16 **	−2.69	−0.06	0.23***	−2.94
	(0.05)	(0.05)		(0.09)	(0.04)
Other race/ethnicity	0.12	0.13		−0.48**	−0.08
Other race/ethnicity	(0.19)	(0.08)		(0.16)	(0.08)
Age ≤ 20	0.09	−0.03		0.04	0.11***
	(0.06)	(0.03)		(0.07)	(0.03)
Age 30-49	−0.05	−0.01		0.11	0.09**
	(0.04)	(0.03)		(0.05)	(0.03)
Age ≥ 50	−0.20	−0.16		−0.35***	−0.14
	(0.12)	(0.09)		(0.11)	(0.07)
Inverse Mills ratio	−0.90***	−0.50		−0.97***	−1.07***
Inverse Mills ratio	(0.25)	(0.24)		(0.29)	(0.20)
χ²	2,471.72	19,287.63		1,230.66	3,379.70
N ₁	7,478	40,445		7,478	40,445

Note. Models also include six dummy variables for the years defendants were processed. Hierarchical linear model predicting bail amount. Hierarchical Bernoulli model predicting held on bail. Italicized coefficients indicate that the effect varies across counties.

p ≤ .01. ***p ≤ .001.

Table 5.

County-Level Effects on Pretrial Release Decisions and Outcomes (Level 1 Slopes and Intercepts As Outcomes).

	Denied bail	ROR	Bail amount	Held on bail
Main effects
Level 1 intercept	−1.81	−1.24	8.88	−0.15
Female violent crime rate	0.003***	0.001	−0.001	−0.0004
Female violent crime rate	(0.001)	(0.001)	(0.001)	(0.001)
Household disadvantage	−0.04	0.10	−0.15**	0.07
Household disadvantage	(0.04)	(0.07)	(0.06)	(0.06)
Conservative ideology	0.08	−0.31***	−0.10*	−0.05
Conservative ideology	(0.06)	(0.07)	(0.05)	(0.06)
Jail crowding	−0.08	0.73***	−0.21	0.83**
	(0.23)	(0.19)	(0.21)	(0.33)
χ²	4404.19	10986.32	21627.53	5068.15
Cross-level interaction effects
Level 1 coefficient for female	—	0.27***	−0.30***	−0.08*
Female violent crime rate		0.00003	0.001*	0.0001
Female violent crime rate		(0.0004)	(0.0003)	(0.0005)
Household disadvantage	—	−0.03	−0.04*	0.05*
Household disadvantage	—	(0.02)	(0.02)	(0.02)
Conservative ideology	—	−0.01	0.07**	−0.07**
Conservative ideology	—	(0.02)	(0.03)	(0.02)
Jail crowding	—	−0.08	−0.10	0.13
	—	(0.07)	(0.07)	(0.08)
χ²	—	10722.85	22138.98	5040.58
N ₂	62

Note. ROR = release on recognizance. Models also included all the defendant-level variables shown in Tables 2, 3, and 4.

p ≤ .05. **p ≤ .01. ***p ≤ .001.

Main Effects of Defendants’ Sex on Pretrial Decisions and Outcomes

Table 2 shows that, consistent with our first hypothesis, female defendants were treated more leniently throughout the pretrial process. Women were less likely to be denied bail, more likely to be granted ROR, and were given lower bail amounts. Compared to male defendants, women had a 1% lower probability of being denied bail, and 4% higher odds of being granted ROR.⁸ Female defendants also received bail amounts that were 29% lower than those received by men. Inconsistent with Demuth and Steffensmeier’s (2004) finding, defendant’s sex did not affect the odds of being held on bail. However, female defendants typically received lower bail amounts, and so defendants’ sex may have indirectly affected whether they were held on bail. Taken together, these results are consistent with the tenets of focal concerns theory and are generally consistent with Demuth and Steffensmeier’s findings derived from their analyses of earlier waves of the SCPS.

Regarding the other variables in the models, legal variables were particularly salient in initial detention decisions, bail amount decisions, and whether defendants were held on bail. For example, judges typically treated defendants arrested for more serious offenses (e.g., murder, robbery, assault, other violent offenses) and with more extensive criminal histories (e.g., prior felony convictions, active criminal justice statuses) more harshly than other defendants. These results are consistent with extant research on court processing; legal factors strongly influence court actors’ social control decisions (e.g., Hartley et al., 2007; Johnson, 2006; Johnson et al., 2008; Savelsberg, 1992; Ulmer, 2012). These findings are also supportive of the focal concerns perspective (Steffensmeier et al., 1998). Judges may perceive defendants with more extensive criminal histories and those charged with more serious offenses as more blameworthy, and thus treat these defendants more harshly. These decisions may also reflect judicial efforts to protect the community from defendants they perceive as more dangerous. Given that these defendants have more at stake (e.g., harsher sentencing), they may be perceived as a greater flight risk to avoid punishment.

Extralegal factors (other than sex) also influenced pretrial decisions and outcomes, albeit to a lesser degree than legal case characteristics. Specifically, ethnicity and sex appear to have influenced judicial perceptions of culpability and the risk of reoffending. Hispanic defendants were treated more punitively across all stages of the pretrial process (with the exception of the ROR decision). This may be due to judicial perceptions that, compared with non-Hispanic defendants, Hispanics are greater flight risks because of their citizenship status; judges may assume that Hispanic defendants are more likely to be noncitizens (Demuth, 2003; see also Hartley et al., 2007). Judges may also be influenced by an increasing number of negative stereotypes that portray Hispanics as lazy, irresponsible, and highly integrated into criminal cultures (Marin, 1984; Portillos, 1998). Defendants’ age also affected pretrial outcomes, although age effects were not consistent across all decision points.

Interaction Effects of Sex and Other Defendant/Case Characteristics on Pretrial Decisions and Outcomes

The analyses of the female defendant samples (Tables 3 and 4) show that some women were treated differently than others during the pretrial process. In general, women arrested for more serious offenses were more likely to be denied bail, less likely to be granted ROR, and received higher bail amounts than women arrested for property offenses. Overall, however, they were not more likely to be held on bail. Furthermore, women with more extensive criminal histories were more likely to be denied bail and less likely to be granted ROR than women with less extensive criminal histories. Criminal history did not influence judges’ bail amount decisions for female defendants, but did increase the likelihood that a woman would be held on bail. Finally, female defendants who had higher bail amounts imposed in their cases were more likely to be held on bail than women afforded lower bail amounts. Overall, these findings support our second hypothesis and suggest that women arrested for more serious offenses and women with lengthier prior records are considered more blameworthy or a greater threat to the community, compared with other women (Steffensmeier et al., 1998). Judges may also perceive these women to be nonconforming to their traditional female role (Chesney-Lind, 1989; Kruttschnitt, 1981), and treat them more harshly in an effort to reinforce these roles.

Demographic characteristics also predicted differential treatment among female defendants, albeit to a lesser degree than legal variables. African American women were less likely to be denied bail than White women, and when a financial release was imposed, they were less likely to post bail. Women who were a race/ethnicity other than White, African American, or Hispanic were less likely to be held on bail. Finally, women younger than 21 and women older than age 49 were less likely to be denied bail and less likely to be held on bail, respectively, compared with women of ages 21 to 29. The largely null impact of extralegal factors does not support our second hypothesis. These findings suggest that judges primarily consider legally relevant criteria when they process women during pretrial and may simply be unwilling to offer additional leniency to female defendants based on their other personal attributes.

The equality of coefficients tests revealed that legal and extralegal factors differentially influenced pretrial decisions and outcomes for female versus male defendants. Specifically, an arrest for robbery or drug trafficking was more influential on judges’ decisions to deny bail and grant ROR among women versus men. Furthermore, an arrest for a drug offense (other than trafficking) and a more extensive criminal history had a greater impact on judicial decisions to deny bail to women than these factors did among men. An arrest for murder more strongly influenced decisions whether to grant ROR among female than male defendants, while an arrest for a violent offense other than murder, robbery, or assault was a more salient predictor of bail amount decisions among male than female defendants. Differences in the magnitude of the effects that impacted who was held on bail included the stronger positive effects of robbery, prior felony conviction, and prior failure to appear among women and the stronger inverse effects for assault, drug trafficking and other drug offense, and public order offense for men.

Our findings that the magnitude of effects of a number of legal factors on judges’ initial detention decisions differed based on defendants’ sex are contrary to our third hypothesis and focal concerns theory, but more consistent with the tenets of the evil woman hypothesis. In particular, deviating from traditionally prescribed gender roles and engaging in more serious criminal behavior or more criminal behavior strongly impacted punitive pretrial decisions and outcomes among women, while these factors were less relevant among men. Judges may perceive that women with such characteristics are particularly atypical relative to other females that come to the attention of the justice system, as these women comprise a lower proportion of the female defendant population compared with the higher density of men with these characteristics in the male defendant population.

In line with the evil woman perspective, it may be that judges identify these characteristics as evidence that such woman should not be privy to the leniency normally granted to female defendants. Although it is possible that such women may still be treated more leniently than men, the gender gap in judicial treatment is reduced. It is unlikely that judicial decisions to weigh certain factors more heavily for women than men stems from beliefs that these female defendants are more dangerous or more likely to reoffend; there is virtually no empirical evidence that indicates this is the case. In fact, the onset and desistance from violence occur earlier for females compared with males, and men typically offend at higher rates than women (Steffensmeier & Allan, 1996). Thus, it simply appears that judges grant females a degree of leniency until their behaviors are seen as no longer deserving of this protection. Once this threshold is reached, judges weigh certain indicators of offense severity and criminal history more strongly among females than males. Men do not appear to be afforded the same leniency because judges do not perceive these behaviors as atypical of the majority of men brought to the attention of the courts. Contrary to this perspective, however, we found few differences with regard to defendants’ sex conditioning the effects of legal factors on bail amounts, which would refute the evil women hypothesis and lend support to the focal concerns perspective. That is, any defendant, regardless of sex, who commits more serious offenses and is considered to be more blameworthy and more of a risk to the community is dealt with more punitively (in terms of bail amount).

Extralegal factors were less influential in explaining pretrial decisions and outcomes for females compared to males, although very few of these effects differed in magnitude between the sexes. The only differences that did emerge were the stronger inverse effects of race (African American) and age (≤20) on judges’ decisions to deny bail among women and the stronger positive effects of ethnicity (Hispanic) on judges’ bail decisions and the odds of being held on bail among men. For the most part, however, these findings refute our fourth hypothesis. Taken together, the comparisons between the sex-specific samples suggest that both legal and extralegal factors differentially affect judicial treatment of female defendants versus males during pretrial, but these differences were more evident during initial detention decisions versus bail decisions, and more salient for legal factors than extralegal factors.

County Contextual Effects on Pretrial Decisions and Outcomes

Results from the unconditional models (not shown) revealed that all the outcomes varied across counties (p ≤ .05), permitting examination of the main effects of the county contextual factors on the Level 1 intercepts. The random coefficients models also revealed that the Level 1 relationships between defendants’ sex and ROR, bail amount, and held on bail each varied across counties (Table 2); these findings partially support our fifth hypothesis and also allowed for examination of the cross-level interactions between the county contextual factors and the Level 1 slopes depicting those relationships.⁹

The main effects reported in Table 5 show that judges were more likely to deny defendants bail in counties with higher female violent crime rates; however, female crime rate did not impact any other pretrial decisions or outcomes. In more conservative counties, judges were less likely to grant defendants ROR, but when financial releases were given, bail amounts were lower in more conservative counties. In counties with less available jail space, judges granted ROR more frequently, but defendants were also more likely to be held on bail when given a financial release. Finally, judges imposed lower bail amounts in counties with higher levels of household disadvantage. Taken together, the main effects of the county contextual factors support the notion that judges are influenced by the sociopolitical conditions within their court community (e.g., Eisenstein et al., 1988; Helms & Jacobs, 2002; Johnson, 2006; Ulmer & Johnson, 2004).

The analyses of the county contextual factors on the Level 1 slopes depicting the relationship between defendants’ sex and ROR revealed no significant cross-level effects. That is, none of the county contextual factors examined here moderated judges’ decisions to grant female defendants ROR.¹⁰ The inverse relationship between defendants’ sex and bail amount was weaker in counties with higher female violent crime rates and in counties with more conservative ideologies. The inverse Level 1 relationship between defendants’ sex and bail amount was stronger in counties with higher levels of household disadvantage. Jail crowding did not moderate the relationship between sex and pretrial decisions and outcomes. The inverse relationship between defendants’ sex and held on bail was weaker in counties with higher levels of household disadvantage, but stronger in counties with more conservative ideologies. Altogether then, the findings pertaining to the moderating effects of county context on the relationship between defendants’ sex and pretrial outcomes offered limited support for Hypotheses 5a, 5b, 5c, and 5d.

The moderating effects of the county contextual factors on the defendants’ sex–bail amount relationship suggest that there may be going rates for bail amounts in these counties, and that the going rates for female defendants may indeed be shaped by the broader sociopolitical context in which these court communities are situated. The effect of female violent crime rate on the defendants’ sex–bail amount relationship underscores that judges are aware of crime rates in surrounding areas and aim to protect their communities. Female defendants who come to the attention of courts in areas with higher rates of female violent crime may be viewed as a greater risk to those communities, which could contribute to judges’ decisions to impose higher bail amounts for these women and bail amounts that are more similar to those imposed for men. In contrast, judges imposed lower bail amounts for women processed in counties with higher levels of household disadvantage, which also suggests that judges are cognizant of the social conditions in the counties in which their courts are situated and consider the practical and social costs of their decisions. Judges may not want to contribute to even higher levels of broken homes or single-parent families in these counties. Although areas with higher levels of structural disadvantage may evoke judicial perceptions of a breakdown in informal social controls, and in general, a response that imposes greater formal control over defendants (see, for example, Britt, 2000; Sampson & Laub, 1993), judges may treat women with greater leniency in disadvantaged counties because they view women as caregivers, who have the capacity to provide informal control over children (Daly, 1987, 1989; Koons-Witt, 2002). Judges may also consider female defendants less blameworthy in counties with higher household disadvantage because they attribute blame for their offending to environmental circumstances such as economic marginalization or familial instability. Judges may treat women more harshly in counties that are more conservative, however, because they are influenced by the prevailing external pressures in their court community. Conservative ideologies tend to be more punitive (Unnever & Cullen, 2006; Unnever et al., 2005), and so it could be that female defendants processed in more conservative counties are simply punished more harshly relative to other women.

Conclusion

In this study, we examined whether judicial decisions made during pretrial generated sex-based disparities as well as whether defendants’ sex interacted with other defendant or case characteristics to produce disparities. We also assessed whether judges’ treatment of female defendants varied across counties; and if so, whether contextual characteristics of counties conditioned the relationship between defendants’ sex and pretrial decisions and outcomes.

Consistent with our prediction, we found evidence that, overall, judges treated female defendants more leniently than males throughout the pretrial process. We also observed that not all female defendants were treated equally—a notion consistent with focal concerns theory (see Spohn & Holleran, 2000; Steffensmeier et al., 1998). Specifically, female defendants arrested for more serious offenses and those with lengthier prior records were treated more harshly than other women, suggesting that judges may have been guided by their focal concerns related to defendants’ blameworthiness and a desire to protect the community. These findings were also consistent with our prediction. Virtually no differences were found, however, across groups of female defendants defined by extralegal or symbolic markers, such as age and race. Judges consistently treated female defendants more leniently, but that was the extent of their leniency. These latter findings did not support our second hypothesis or the focal concerns perspective, although they are consistent with findings from some research on sentencing outcomes (e.g., Spohn & Beichner, 2000; Steffensmeier & Demuth, 2006).

We also found evidence that some legal factors were more influential on judges’ pretrial decisions among female versus male defendants. A number of legal factors that could be considered more “masculine” in nature (e.g., committing a violent offense) were stronger predictors of more punitive pretrial outcomes among women. While these findings were inconsistent with our expectations and with the tenets of the focal concerns perspective, they were consistent with the evil woman hypothesis, as they might reflect judicial attempts to reinforce traditional gender roles through legal sanctions (Chesney-Lind, 1989; Kruttschnitt, 1981). Also inconsistent with our predictions and the focal concerns perspective, extralegal factors were, for the most part, not influential in explaining pretrial outcomes among females or males, nor did the magnitude of these effects vary considerably between the sexes.

Court researchers have revealed important variation between counties in judicial sentencing decisions (e.g., Johnson, 2006; Ulmer & Johnson, 2004). We extended this line of research to judicial decisions made during pretrial and observed that between-county variation in decision making also exists during this earlier stage of the process, and the sociopolitical context of the county shapes these decisions. With regard to the treatment of women during pretrial, however, we found that the county contextual factors examined here only impacted judicial decisions in a limited capacity. County contextual factors did not impact the between-county variability in the treatment of women during the initial stages of the pretrial process, but the county contextual factors did explain some of the variability in judges’ bail amount decisions.

The defendant-level findings and the findings regarding the direct and moderating influence of county context discussed above are important and have intuitive appeal; however, they are also only generalizeable to courts operating in the large urban counties in the United States. Scholars have suggested that the size and location of the court may influence case processing (see, for example, Eisenstein et al., 1988). An important avenue for future inquiry could be to examine whether the findings we observed in this study hold for courts located in rural or less urban areas. In addition, the SCPS is a rich source of data, but it does not include several measures that could, if included in our models, confound our results. Researchers may, therefore, wish to expand on our findings by examining the impact of other potentially relevant predictors of judicial decisions that are not included in the SCPS data set. At the defendant level, researchers might include measures of defendants’ social class, income, employment status, length of residence in the community, number of dependent children, and attorney type. Future research may also consider examining more nuanced measures of offense seriousness, as there is potentially relevant variation within the offense categories included here that may coincide with sex-based differences in judicial treatment. Furthermore, it is also possible that our findings are biased because we did not include a measure of whether the jurisdictions included here operated under bail guidelines that structure judicial decision making during pretrial. Although we included many of the defendant-level factors often used as markers in bail guidelines in our analyses, future research might examine the impact of bail guidelines on pretrial decision making.

The limitations of this study and related areas for future study aside, our findings underscore the need for future research on pretrial and other understudied stages of the criminal court process. Recall, decisions during pretrial must balance the goal of preserving the defendant’s liberty with the goal of protecting the community, and some researchers have found that such decisions may affect later court processing (Albonetti, 1991; Griffin & Wooldredge, 2006; Lizotte, 1978; Spohn, 2009). It is only through continued examination of all stages of the criminal court process that we can better understand the ways in which defendants are processed in this nation’s criminal 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) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Gillian M. Pinchevsky is a faculty member in the Department of Criminal Justice at the University of Nevada, Las Vegas.

Benjamin Steiner is an assistant professor in the School of Criminology and Criminal Justice at the University of Nebraska at Omaha. He holds a PhD from the University of Cincinnati.

References

Alarid

L. F.

Marquart

J. W.

Burton

V. S.

Jr Cullen

F. T.

Cuvelier

S. J.

(1996). Women’s roles in serious offenses: A study of adult felons. Justice Quarterly, 13, 431-454.

Albonetti

C. A.

(1986). Criminality, prosecutorial screening, and uncertainty: Toward a theory of discretionary decision making in felony case processings. Criminology, 24, 623-644.

Albonetti

C. A.

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

Albonetti

C. A.

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

Applegate

B. K.

Cullen

F. T.

Fisher

B. S.

Vander Ven

(2000). Forgiveness and fundamentalism: Reconsidering the relationship between correctional attitudes and religion. Criminology, 38, 719-753.

Brame

Paternoster

Mazerolle

Piquero

(1998). Testing for the equality of maximum-likelihood regression coefficients between two independent equations. Journal of Quantitative Criminology, 14, 245-261.

Bridges

G. S.

Steen

(1998). Racial disparities in official assessments of juvenile offenders: Attributional stereotypes as mediating mechanisms. American Sociological Review, 63, 554-570.

Britt

(2000). Social context and racial disparities in punishment decisions. Justice Quarterly, 17, 707-732.

Browne

Miller

Maguin

(1999). Prevalence and severity of lifetime physical and sexual victimization among incarcerated women. International Journal of Law and Psychiatry, 22, 301-322.

10.

Bushway

S. D.

Johnson

B. D.

Slocum

L. A.

(2007). Is the magic still there? The use of the Heckman two-step correction for selection bias in criminology. Journal of Quantitative Criminology, 23, 151-178.

11.

Bushway

S. D.

Piehl

A. M.

(2001). Judging judicial discretion: Legal factors and racial discrimination in sentencing. Law & Society Review, 35, 733-764.

12.

Chesney-Lind

(1989). Girls’ crime and woman’s place: Toward a feminist model of female delinquency. Crime & Delinquency, 35, 5-29.

13.

Clear

T. R.

(2007). Imprisoning communities: How mass incarceration makes disadvantaged neighborhoods worse. Oxford, UK: Oxford University Press.

14.

Clogg

C.C.

Petkova

Haritou

(1995). Statistical methods for comparing regression coefficients between models. American Journal of Sociology, 100, 1261-1293.

15.

Cohen

T. H.

Kyckelhahn

(2010). Felony defendants in large urban counties, 2006. Washington, DC: Bureau of Justice Statistics.

16.

Daly

(1987). Discrimination in the criminal courts: Family, gender, and the problem of equal treatment. Social Forces, 66, 152-175.

17.

Daly

(1989). Neither conflict nor labeling nor paternalism will suffice: Intersections of race, ethnicity, gender, and family in criminal court decisions. Crime & Delinquency, 35, 136-168.

18.

Daly

Bordt

R. L.

(1995). Sex effects and sentencing: An analysis of the statistical literature. Justice Quarterly, 12, 142-175.

19.

Daly

Tonry

(1997). Gender, race, and sentencing. Crime and Justice, 22, 201-252.

20.

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.

21.

Demuth

Steffensmeier

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

22.

Dubin

J. A.

Rivers

(1990). Selection bias in linear regression, logit and probit models. In Fox

Long

J. S.

(Eds.), Modern methods of data analysis (pp. 410-442). Newbury Park, CA: SAGE.

23.

Eisenstein

Flemming

R. B.

Nardulli

P. F.

(1988). The contours of justice: Communities and their courts. Lanham, MD: University Press of America.

24.

Fearn

N. E.

(2005). A multilevel analysis of community effects on criminal sentencing. Justice Quarterly, 22, 452-487.

25.

Goldkamp

J. S.

(1979). Two classes of accused: A study of bail and detention in American justice. Cambridge, MA: Ballinger.

26.

Gottfredson

M. R.

Gottfredson

D. M.

(1988). Decision making in criminal justice: Toward the rational exercise of discretion (2nd ed.). New York, NY: Plenum.

27.

Greene

(2005). Censored data and truncated distributions. In Mills

T. C.

Patterson

(Eds.), The handbook of econometrics (Vol. 1, pp. 695-735). London, England: Palgrave MacMillan.

28.

Griffin

Wooldredge

(2006). Sex-based disparities in felony dispositions before versus after sentencing reform in Ohio. Criminology, 44, 893-923.

29.

Hanushek

E. A.

Jackson

J. E.

(1977). Statistical methods for social scientists. New York, NY: Academic Press.

30.

Harlow

C. W.

(1999). Prior abuse reported by inmates and probationers. Washington, DC: U.S. Department of Justice.

31.

Hartley

R. D.

Maddan

Spohn

C. C.

(2007). Concerning conceptualization and operationalization: Sentencing data and the focal concerns perspective—A research note. Southwest Journal of Criminal Justice, 4, 58-78.

32.

Hawkins

(1987). Beyond anomalies: Rethinking the conflict perspective on race and criminal punishment. Social Forces, 65, 719-745.

33.

Heckman

J. J.

(1976). The common structure of statistical models of truncation, sample selection and limited dependent variables and a simple estimator for such models. Annals of Economic and Social Measurement, 5, 475-492.

34.

Helms

Jacobs

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

35.

Johnson

B. D.

(2003). Racial and ethnic disparities in sentencing departures across modes of conviction. Criminology, 41, 449-489.

36.

Johnson

B. D.

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

37.

Johnson

B. D.

Ulmer

J. T.

Kramer

J. H.

(2008). The social context of guidelines circumvention: The case of federal district courts. Criminology, 46, 737-783.

38.

Koons-Witt

B. A.

(2002). The effect of gender on the decision to incarcerate before and after the introduction of sentencing guidelines. Criminology, 40, 297-328.

39.

Kruttschnitt

(1981). Social statuses and sentences of female offenders. Law & Society Review, 15, 247-265.

40.

Kruttschnitt

McCarthy

(1985). Familial social control and pretrial sanctions: Does sex really matter? Journal of Criminal Law and Criminology, 76, 151-175.

41.

Lizotte

A. J.

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

42.

Marin

(1984). Stereotyping Hispanics: The differential effect of research method, label, and degree of contact. International Journal of Intercultural Relations, 8, 17-27.

43.

Mitchell

(2005). A meta-analysis of race and sentencing research: Explaining the inconsistencies. Journal of Quantitative Criminology, 21, 439-466.

44.

Mullins

C. W.

Wright

(2003). Gender, social networks, and residential burglary. Criminology, 41, 813-839.

45.

Myers

M. A.

Talarico

S. M.

(1987). The social contexts of criminal sentencing. New York, NY: Springer-Verlag.

46.

Nagel

I. H.

Hagan

(1983). Gender and crime: Offense patterns and criminal court sanctions. Crime and Justice, 4, 91-144.

47.

Payne

B. K.

Gainey

R. R.

Triplett

R. A.

Danner

M. J. E.

(2004). What drives punitive beliefs? Demographic characteristics and justifications for sentencing. Journal of Criminal Justice, 32, 195-206.

48.

Portillos

E. L.

(1998). Latinos and Latinas: Latinos, gangs, and drugs. In Mann

Zatz

M. S.

(Eds.), Images of color, images of crime. Los Angeles, CA: Roxbury.

49.

Raudenbush

S. W.

Bryk

A. S.

Cheong

Y. F.

Congdon

(2011). HLM: Hierarchical linear and nonlinear modeling (7.0 ed.). Lincolnwood, IL: Scientific Software Internatonal, Inc.

50.

Sampson

R. J.

Groves

W. B.

(1989). Community structure and crime: Testing social-disorganization theory. American Journal of Sociology, 94, 774-802.

51.

Sampson

R. J.

Laub

J. H.

(1993). Structural variations in juvenile court processing: Inequality, the underclass, and social control. Law & Society Review, 27, 285-311.

52.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918-924.

53.

Savelsberg

J. J.

(1992). Law that does not fit society: Sentencing guidelines as a neoclassical reaction to the dilemmas of substantivized law. American Journal of Sociology, 97, 1346-1381.

54.

Spohn

(2000). Thirty years of sentencing reform: The quest for a racially neutral sentencing process. In National Institute of Justice (Ed.), Criminal Justice 2000: Politics, Processes, and Decisions of the Criminal Justice System (pp. 427-501). Washington, DC: National Institute of Justice/National Criminal Justice Reference Service.

55.

Spohn

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

56.

Spohn

Beichner

(2000). Is preferential treatment of female offenders a thing of the past? A multisite study of gender, race, and imprisonment. Criminal Justice Policy Review, 11, 149-184.

57.

Spohn

Holleran

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

58.

Steen

Engen

R. L.

Gainey

R. R.

(2005). Images of danger and culpability: Racial stereotyping, case processing, and criminal sentencing. Criminology, 43, 435-468.

59.

Steffensmeier

Allan

(1996). Gender and crime: Toward a gendered theory of female offending. Annual Review of Sociology, 22, 459-487.

60.

Steffensmeier

Demuth

(2006). Does gender modify the effects of race-ethnicity on criminal sanctioning? Sentences for male and female White, Black, and Hispanic defendants. Journal of Quantitative Criminology, 22, 241-261.

61.

Steffensmeier

Kramer

Streifel

(1993). Gender and imprisonment decisions. Criminology, 31, 411-446.

62.

Steffensmeier

Terry

R. M.

(1986). Institutional sexism in the underworld: A view from the inside. Sociological Inquiry, 56, 304-323.

63.

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.

64.

Ulmer

J. T.

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

65.

Ulmer

J. T.

Bader

Gault

(2008). Do moral communities play a role in criminal sentencing? Evidence from Pennsylvania. Sociological Quarterly, 49, 737-768.

66.

Ulmer

J. T.

Bradley

M. S.

(2006). Variation in trial penalties among serious violent offenses. Criminology, 44, 631-670.

67.

Ulmer

J. T.

Johnson

(2004). Sentencing in context: A multilevel analysis. Criminology, 42, 137-177.

68.

Unnever

Cullen

F. T.

(2006). Christian fundamentalism and support for capital punishment. Journal of Research in Crime & Delinquency, 43, 169-197.

69.

Unnever

Cullen

F. T.

Applegate

B. K.

(2005). Turning the other cheek: Reassessing the impact of religion on punitive ideology. Justice Quarterly, 22, 304-339.

70.

Walker

(1993). Taming the system: The control of discretion in criminal justice, 1950-1990. New York, NY: Oxford University Press.

71.

Wooldredge

Griffin

Rauschenberg

(2005). (Un)anticipated effects of sentencing reform on the disparate treatment of defendants. Law & Society Review, 39, 835-873.

72.

Zatz

M. S.

(2000). The convergence of race, ethnicity, gender, and class on court decision making: Looking toward the 21st century. In National Institute of Justice (Ed.), Criminal Justice 2000: Politics, Processes, and Decisions of the Criminal Justice System (pp. 503-552). Washington, DC: National Institute of Justice/National Criminal Justice Reference Service.