Assessing Cumulative Disadvantage against Minority Female Defendants in State Courts

Abstract

Prior sentencing research, especially research on cumulative disadvantage, has mainly focused on the treatment of male defendants. Little attention has been paid to female defendants, particularly minority female defendants. Drawing on the selective chivalry, evil women, and focal concerns perspectives and using data from the 1990–2009 State Court Processing Statistics (SCPS), this paper investigates the impact of race/ethnicity for female defendants across individual and successive stages in the sentencing process. The results indicate that ethnicity does not operate via indirect or direct pathways, and therefore no evidence of cumulative disadvantage against Hispanic female defendants was detected. The results, however, do suggest that race operates through direct and indirect pathways to cause more punitive sentencing outcomes for Black female defendants compared to White female defendants, thus providing evidence of cumulative disadvantage against Black female defendants. Theoretical, research, and policy implications are discussed.

Keywords:

• Gender
• race/ethnicity
• cumulative disadvantage
• sentencing disparity

Acknowledgements

We want to thank Cassia Spohn and Cody Telep for their helpful comments and suggestions. We also want to thank Jeffery Ulmer and anonymous reviewers for their constructive feedback and insights.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 While the 1990–2009 SCPS data are well suited for this study and contain pertinent measures to address our research question, these data may not necessarily be reflective of current practices in state court processing. Therefore, more recent data, when available, may be used to replicate and test for cumulative disadvantage against minority female defendants in state courts.

2 By using these relatively limited criminal history measures, we may have grouped defendants together into the same category when they in fact possess substantively different backgrounds, “which may cloud our understanding of this variable and how it is used in courtroom decision making” (Tillyer et al., 2015, p. 718). This may be problematic because the more lenient treatment white female defendants receive may depend on their criminal histories as women with limited criminal histories may be viewed as conforming to traditional gender roles but women with substantial criminal histories may be seen as violating traditional gender roles (see Tillyer et al., 2015). Future research may want to explore the use of a most nuanced variable of criminal history and assess if the more lenient treatment white female defendants receive indeed depends on their criminal histories.

3 Generally, dismissals refer to an order or judgment, only by a judge, in which the case is disposed of without a trial, whereas acquittals refer to defendants exonerated (found not guilty) on charges. Acquittals are generally more serious than dismissals, as the defendant must go through a formal trial process (Goldkamp, 1980). However, the SCPS data do not separate dismissals from acquittals in this variable. Future research may want to consider separately examining dismissals and acquittals.

4 We use the maximum sentence length available in the data (840 months) for defendants receiving life imprisonment or death penalty. Thus, cases where the defendant received life imprisonment (N = 6) are coded as the maximum prison length of 840 months, and there were no women sentenced to the death penalty in the SCPS data.

5 Some scholars also include prior failure to appear (FTA) as a measure of criminal history, which identifies whether the offender had a history of FTA for prior court dispositions (see Demuth, 2003; Franklin & Fearn, 2015). This variable, however, was highly correlated with pretrial detention (r = .96, p < .01). Thus, we did not include FTA in the analysis.

6 Our offense severity measures—that is, a dummy variable for multiple arrest charges and a series of 15 dummy variables that identify the most severe type of charge the defendant was arrested on—may not truly capture differences in the severity of offenses that defendants of different backgrounds commit. It may be that Black women commit more or less serious forms of assault, for example, than White women. Our measures of offense severity, however, may not be able to capture this. Unfortunately, the SCPS data do not provide any other way to assess offense severity.

7 A conviction charge variable is often used in sentencing models to account for the offense severity. The conviction charge variable in the SCPS data contains 17 categories, including a misdemeanor charge and 16 most severe felony offenses defendants were charged with. This variable thus overlaps with adjudication (1 = felony; 0 = misdemeanor), causing a harmful level of multicollinearity. Because this study focuses on multiple decision points and cumulative disadvantage through successive stages, we include the adjudication variable, which allows us to assess the effects of race/ethnicity on sentencing through adjudication. Thus, we could not control for conviction charge severity; instead, we control for arrest charge severity in the decision to incarcerate and sentence length models (see Tables 5 and 6).

8 Specifically, we included a series of dummy variables to capture any potentially important inter-county variations at each decision-making stage. Because this paper focuses on individual-level racial disparities among female defendants, this analytic approach is appropriate and parsimonious (Johnson & Betsinger, 2009, p. 1063). Thus, we did not explore the specifics of whether the significant findings related to race/ethnicity, or the lack of those, are equally robust across different counties. It is possible that the effects of race/ethnicity may be more pronounced in counties that place a greater emphasis on conforming to traditional gender roles, but may be attenuated in counties that adopt less traditional expectations for females (see Kim et al., 2019). Future research thus may want to assess if cumulative disadvantage against minority female defendants is amplified or reduced by relevant county-level measures, such as political context.

9 It can be argued that the decision to incarcerate is an ordinal variable, and thus the use of ordinal logistic regression is appropriate. However, ordinal logistic regression requires the proportional odds assumption be met, and the Brant test conducted in Stata 15 indicated that this assumption was violated.

10 In traditional path analysis, one can take each OLS estimate that compromises a path/paths to determine the size of an indirect effect on the last outcome (see Brennan, 2006). However, because our analysis uses binary and multinomial logistic regressions for several outcomes, traditional path analysis is not appropriate. Instead, gsem is used to specify the outcomes of interest appropriately.

11 The sample size is reduced from the originally reported (N = 18,898) to the analysis of pretrial detention (N = 15,051) when listwise deletion is used. Further reduction of the sample size across case processing decisions is due to the fact that pretrial detention, adjudication, the decision to incarcerate, and sentence length are all a part of process in which some cases will not reach these outcomes because they are not adjudicated or sentenced. As the number of cases is reduced from pretrial detention to adjudication and then sentencing, there may be inherent differences between the cases that do and do not make it to sentencing. Notably, Heckman’s correction is often used to account for this potential selection bias. However, as Bushway, Johnson, and Slocum (2007, pp. 161–162) noted, “The Heckman two-step estimator is specifically a probit model followed by a linear regression, and there is no simple analog of the Heckman method for discrete choice models despite the logical appeal of the process” (see also Dubin & Rivers, 1990). Since our incarceration decision variable is a multinomial outcome, it would not be appropriate to predict incarceration decision using OLS. Thus, it would not be appropriate to calculate a Heckman adjustment in the incarceration decision model, which requires the use of probit to predict adjudication and OLS to predict incarceration decision. Further, since cases in which the defendant did not receive a term of incarceration were coded as 0 months in this study (see Kim et al., 2019; Starr & Rehavi, 2013), all defendants that were sentenced were included in the analysis of sentence length. As such, no Heckman correction is necessary for the analysis of sentence length.

12 Despite that the focus of this study is on female defendants, we want to provide a backdrop to understand the experience of all race/ethnicity/gender groups by contextualizing the experience of minority female defendants in relation to the experience of all defendants at each individual decision-making stage. To this end, we created six dummies representing Black female, Hispanic female, White female, Black male, Hispanic male, and White male defendants, respectively. We then assessed how minority female defendants were treated compared to the male group subject to the most severe treatment and the male group subject to the least severe treatment, respectively, by using Black male defendants as the reference category (see Appendix A) and White male defendants as the reference category (see Appendix B), respectively. The results suggest that when compared to Black male defendants, all female defendants, regardless of race/ethnicity, benefit from chivalry by receiving more lenient treatment in pretrial detention, incarceration decision, and sentence length. When compared to White male defendants, however, race/ethnicity matters. Specifically, whereas White female defendants benefit from chivalry in pretrial detention, prison sentences, and sentence length by receiving more lenient treatment than White male defendants in these three decisions, Hispanic female defendants do not benefit from chivalry in any decision and Black female defendants only benefit from chivalry in prison sentences and sentence length by receiving more lenient treatment than White male defendants in these decision. Overall, these findings provide support for the selective chivalry and evil women perspectives.

13 The valid number of cases for each outcome across race/ethnicity is as follows. Pretrial Detention: Black women (N = 7,525); Hispanic women (N = 1,121); White women (N = 6,405). Dismissal/Acquittal: Black women (N = 5,768); Hispanic women (N = 814); White women (N = 4,922). Felony Adjudication: Black women (N = 4,062); Hispanic women (N = 533); White women (N = 3,685). The Decision to Incarcerate and Sentence Length have the same number of cases: Black women (N = 3,799); Hispanic women (N = 462); White women (N = 3,489).

14 Regression coefficients in models with logged dependent variables—in this case, sentence length—are often directly interpreted as the percent change in the dependent variable without any transformation (Johnson & Betsinger, 2009, p. 1062).

15 One anonymous reviewer suggested that given that our Hispanic female sample is considerably smaller than the White and Black samples, there may be some outcomes that have significantly fewer Hispanic female cases, which may also be a reason that few Hispanic effects were statistically significant in the multivariate analyses. We thank the reviewer for suggesting this possible explanation.

Additional information

Notes on contributors

Kelsey L. Kramer

Kelsey L. Kramer, MS, is a doctoral student in the Department of Criminal Justice and Criminology at Sam Houston State University. Her primary research interests include gender and race/ethnicity in criminal justice; disparities in punishment—particularly sentencing; and policing.

Xia Wang

Xia Wang, PhD, is an Associate Professor at Arizona State University’s School of Criminology and Criminal Justice. She is involved in studies of race and ethnicity and their effects on crime and criminal justice, and the use of various analyses to test and extend criminological theories. Her work has appeared in Criminology, Journal of Research in Crime and Delinquency, Journal of Quantitative Criminology, Justice Quarterly, Law & Society Review, and other journals.