The “Distance Traveled”: Investigating the Downstream Consequences of Charge Reductions for Disparities in Incarceration

Abstract

Relatively little work examines the impact that charging decisions exert on sentencing. We investigate this issue by estimating the “distance traveled” in charge bargaining, or the expected change in the likelihood of incarceration associated with reductions in charges across different stages of prosecution. Using data from New York County, we examine how the probability of incarceration shifts as a result of charging decisions and how this potentially contributes to social inequalities in incarceration. Findings indicate that charge reductions are associated with sizeable decreases in the probability of incarceration, particularly at the plea bargaining stage. On average, the “distance traveled” is substantially greater for female than male defendants and for White compared to Latino and Black defendants, even after accounting for a host of relevant punishment factors. Findings are discussed as they relate to contemporary theoretical perspectives on prosecutorial decision-making and social inequality in punishment.

Keywords:

• Distance traveled
• charge bargaining
• guilty pleas
• disparity
• sentencing

Acknowlegement

The authors would like to thank Dr. James Lynch and Dr. Thomas Loughran for their valuable comments on previous drafts of this paper, and they would like to acknowledge helpful research assistance provided by Ms. Christina Stewart.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Brian D. Johnson is Professor of Criminology and Criminal Justice at the University of Maryland. His research interests involve various aspects of social inequality in the justice system, with a particular focus on racial and ethnic disparities in prosecution and punishment.

Pilar Larroulet is a doctoral candidate in the Department of Criminology and Criminal Justice at University of Maryland, College Park, and a professor at Universidad Católica de Chile. Her research interests include developmental and life course criminology and reentry.

Notes

1 Among the felony arrestees convicted by plea, 3.4% of the sample was Asian American and <1% was Native American. A small number of cases were removed from the sample because they were missing necessary information on sentence type (n = 448) or offense type at conviction (n = 859). In addition, 8 cases were removed for missing information on defendant gender along with six cases missing information on neighborhood of arrest.

2 Because we begin with a sample of felony arrests, very few cases are reduced to Class B misdemeanors or violations, particularly at the screening stage (n = 203 and n = 79 cases, respectively). To address concerns with small cell sizes, we collapse misdemeanors and violations into a single “non-felony” classification. Separate analyses separating misdemeanors and violations produce similar results. We also combine Class A and Class B felonies because very few offenses were convicted as Class A felonies (only nine person offenses and 74 drug offenses in our sample).

3 Because the specific types of felony offenses are closely aligned with statutory severity levels (e.g. all first Degree Robberies are Class B Felonies), it is not possible to include both in the model. We therefore examine statutory severity levels along with broader offense categories. Person offenses include crimes such as assault or robbery, property offenses include crimes such as larceny or theft as well as fraud and related offenses, drug offenses include felony possession or distribution, public order includes crimes such as prostitution or gambling, and other offenses involve crimes that do not fit into these other categories, such as obstruction of justice and weapon offenses. Drug crimes are limited to cases prosecuted by DANY. In New York County, drug cases are divided at random between DANY and the Office of the Special Narcotics Prosecutor, with each handling roughly half of all drug cases in the county (see Kutateladze and Andiloro, 2014).

4 We dichotomize this measure because the vast majority of cases are convicted of a single charge (96% in our sample); however, alternative models specifying a continuous measure for the number of charges at conviction produces equivalent results.

5 Supplemental models including the individual criminal history measures separately also produce equivalent findings. These and other supplemental analyses are available from the authors by request.

6 This can occur when a guilty plea or other case disposition is determined prior to arraignment, so that the detention status at arraignment is not applicable or unknown. Reported findings are unaffected by the removal of these cases.

7 In New York county, court-appointed attorneys are referred to as “18(b)” attorneys, pursuant to county law that established them, and public defenders come from several different agencies all of which are nonprofit organizations that provide defense counsel to indigent defendants in different parts of the county. We initially included an additional dummy variable for the small number of cases (n = 605) missing information on type of attorney but they were indistinguishable from the public defender cases and are therefore grouped in the omitted reference category. Results are unchanged when these cases are captured separately using an additional dummy variable.

8 The variance inflation factors (VIFs) for independent variables of interest are all below 3.0. Age and age-squared are highly correlated, but this is to be expected and is not an indicator of problematic multicollinearity, which involves two (or more) variables correlated by chance rather than by design.

9 We do not investigate case acceptance because the overwhelming majority of cases (96%) are forwarded for prosecution in New York county (see Kutateladze et al., 2014).

10 Increases in initial arrest charges can occur for several reasons. Prosecutors may view the defendant or the offense as relatively more serious or more culpable than the police, or additional evidence might come to light after the arrest that strengthens the case against the defendant.

11 Percentage changes are calculated as follows: $\%\Delta = \frac{V_1-V_0}{V_0}\times 100\%$ where V₀=predicted probability of incarceration at Time 1 (arrest or screening) and V₁ is the predicted probability at Time 2 (screening or conviction).

12 Additional analysis of separate binary charge reduction measures at each stage of prosecution (results not shown in tabular form) also demonstrate that key predictors exert different effects across charging decisions. Gender, for example, is significantly and positively related to charge reduction at screening but negatively related to it during plea bargaining, which helps to explain its null effect in the overall binary charge reduction model.

13 We also investigated differences by type of offense for the subset of male offenders (there were insufficient numbers of cases across offense types to conduct this analysis separately for female defendants). Consistent with prior work (e.g. Steffensmeier and Demuth, 2000; Schlesinger, 2005; 2013), racial disparities among men were most pronounced for arrests that involved drug or violent crimes. These additional results are also available by request.

14 Recall that our estimates adjust for all charging changes, including the small number of cases in which charges were increased by prosecutors. The inclusion of all charge alterations is important because it captures the bidirectional processes that shape final case outcomes, but this may also contribute to an underestimate of the effect of charge reductions on sentencing. To investigate this issue, we re-estimated our models after excluding the small number of cases in which the initial charge severity increased, and our overall estimates were little affected. Specifically, the average probability of incarceration at screening and conviction were reduced only slightly from 0.579 to 0.565 and from 0.464 to 0.446, respectively, which is not surprising given the relative rarity of charge increases in our data.

15 Across 25 offense-specific contrasts for gender and race/ethnicity, only two exceptions occurred. At the initial screening phase, males received slightly larger charging discounts for property offenses and Hispanics received the largest discounts for public order crimes. For all other offense-specific comparisons, female and White offenders benefitted most from charge reductions at both the initial screening and plea bargaining stages.