An examination of the direct and interactive effects of race/ethnicity and gender on charge reduction

ABSTRACT

Much of the prior literature on criminal court case processing has focused on judicial decisions regarding bail and sentencing. Fewer studies have examined prosecutorial decision-making, particularly charge reduction. Framed within the focal concerns perspective, this paper examined racial and gender disparity in charge reduction and whether disparity existed across different types of charge reduction. Findings demonstrated partial support for the focal concerns perspective with men and minority defendants less likely to receive a severity reduction.Implications for plea negotiation policies are discussed.

KEYWORDS:

• Charge reduction
• focal concerns
• prosecutorial
• racial disparity

Disclosure of potential conflicts of interest

No potential conflict of interest was reported by the author(s).

Notes

1. We fully acknowledge the complexity of charge reduction and plea negotiation. Prosecutors have multiple tools within their toolkit to induce guilty pleas, such as filing multiple counts or a charge that requires a mandatory minimum for repeat offender status to bring to the table in negotiating a guilty plea. Capturing complexity of charge reduction that may involve decisions to reduce one aspect of a case (i.e., drop counts) as opposed to another (i.e., reduce severity) is difficult to capture in traditional quantitative models (see Bloch, Engen, and Parrotta 2014). Future research with advanced analytical techniques or mixed methods approaches to specific types of cases is needed to fully examine how prosecutors utilize the myriad of tools available during a plea negotiation.

2. In order to have a charge reduction possible from initial charge to conviction, cases would need to be prosecuted to the point of adjudication and result in a guilty plea, rather than going to trial.

3. We also ran the analysis using OLS regression, poisson, and negative binomial models for the amount of a severity reduction and count reduction. The results did not significantly differ from those generated by the logistic regression models we present in the paper (results available from authors upon request).

4. Based upon conversations with both prosecutors and defense attorneys in this jurisdiction, overcharging in the form of stacking counts appears to be relatively uncommon; however, other prosecutor’s offices may have a local culture that supports this (see Ulmer 1997; Johnson 2018). We did not find instances in which prosecutors increased the number of counts between initial filing and adjudication, and only found one instance of an increase in severity. It is likely that the local prosecutor culture does not utilize charge increases as a method of charge negotiation (see Johnson 2018).

5. Although prior literature suggests that evidentiary factors influence charge reduction and plea negotiations, most studies do not include these measures (c.f. Kutateladze, Andiloro, and Johnson 2016). Our data do not have direct measures for strength of evidence, witness credibility, or victim cooperation.

6. The focal concerns perspective suggests that prosecutorial and defense attorney caseload will impact case processing. Unfortunately, we do not have data on caseload for either attorney.

7. We also account for some of the impact of prior decision points and selection bias by modeling case dismissals. To account for the possibility of biased selection into our sample, we estimate a probit model for selection into our sample of plea cases as a function of the information that is available to us in a separate database: defendant race, defendant gender, defendant age, and the crime severity. To satisfy the exclusion restriction, we also include the reviewing District Attorney or Assistant District Attorney. Using the results of the probit model, we calculate the Inverse Mills Ratio (IMR) and re-estimate the models of charge reduction with the IMR included as a control. The results are available in the Supplementary Appendix. We find that the results of our key independent variables of interest are nearly identical to those in the models that do not include an adjustment for sample selection bias. Most importantly, the fundamental inferences we make as a result of the model findings also remain unchanged.

8. We also conducted a series of mediation analyses using both variables as mediators, but the results did not suggest significant mediation effects were occurring and thus were not presented in this paper. Each measure of charge reduction is based on the possibility of a reduction; as such, count reduction includes a subset of cases in which two or more counts were initially charged, violent charge reduction includes only cases which were initially charged as a violent crime.

9. While our ultimate goal is not to explain all variation in charge reductions, we feel some attention should be paid to the measures of fit we provide in Tables 2, 3, and 4. While we include the pseudo R² values, these pseudo R² values are notoriously difficult to interpret in a straightforward manner; therefore, we also calculate the area under the receiver operating characteristic (ROC) curve, which is abbreviated to AUC, as an indicator of model fit for logistic regression. It is equivalent to the probability that a randomly chosen positive instance is ranked higher than a randomly chosen negative instance, i.e. it is equivalent to the two sample Wilcoxon rank-sum statistic.

10. Partial (marginal) effects are calculated using the mfx package in R. Marginal effects use model prediction so we can better interpret the model in the scale that makes more sense.

11. While the effect of a discrete change from Black to White is statistically significant at conventional levels (p < 0.05), the effect of the discrete change from white to Hispanic approaches conventional levels of statistical significance (P < 0.1).

12. Similarly, we find no significant race or gender effects when it comes to the likelihood of count reduction for drug, violent, and all other crimes. Instead, only the total number of counts is associated with the likelihood of a count reduction. Results available upon request.

Additional information

Notes on contributors

Danielle M. Romain Dagenhardt

Dr. Danielle M. Romain Dagenhardt is an assistant professor in the Department of Criminal Justice & Criminology at the University of Wisconsin-Milwaukee.  Her research interests include court processing and gender and racial disparities in court decision-making.  Her work has been published in journals includingFeminist Criminology, Crime & Delinquency, and Journal of Criminal Justice Education.

Amanda J. Heideman

Dr. Amanda Heideman is a Postdoctoral Research Associate in the Department of Criminal Justice & Criminology at the University of Wisconsin-Milwaukee. Her primary research areas are urban politics and policymaking, race/ethnic politics, and issues of race/class/gender inequality in the political and criminal justice systems.

Tina L. Freiburger

Dr. Tina Freiburger is a professor and Helen Bader School of Social Welfare Dean at the University of Wisconsin-Milwaukee. Her primary research areas are courts and sentencing, program evaluation, racial/ethnic issues in the criminal justice system and the intersection of mental health and substance abuse issues in the criminal justice system. Dr. Freiburger’s recent publications include articles inCrime & Delinquency, Feminist Criminology, Criminal Justice Policy Review, and Race and Justice.