Introduction
Extant literature in political science, public policy, psychology, and criminology suggests that the judgement about distributive and procedural fairness, not just effectiveness, is a primary determinant of people’s perception of legitimacy of legal authorities (too voluminous to cite here). For example, Tyler and his colleagues explicitly posit that neutral decision-making and fair treatment of citizens might be as equally as or even more important than doing the job effectively to earn legitimacy, although both are necessary antecedents of citizen’s voluntary compliance with the laws and cooperation with the legal authorities (Sunshine & Tyler, 2003; Tyler, 1990, 2006, 2010, 2017; Tyler & Fagan, 2008; Tyler & Huo, 2002; Tyler & Lind, 1992).
Due to the deep-seated overrepresentation of racial/ethnic minorities observed in a series of criminal decision-making (e.g., pretrial detention, indictment, incarceration), unwarranted racial and ethnic disparity in the decision making of criminal case processing has garnered substantial and continued attention from both scholars and practitioners. The vast majority of evidence suggests that race and ethnicity affect case processing and sentencing outcomes unduly either directly (Donnelly & MacDonald, 2018; Johnson, 2003, 2005, 2006; Kutateladze et al., 2014; Rehavi & Starr, 2014; Sloan, 2022; Tuttle, 2019; Wooldredge et al., 2015), indirectly via earlier decision-makings which are less visible (Kurlychek & Johnson, 2019; Kutateladze et al., 2014; Stolzenberg et al., 2004), or through interactions with other offender/case characteristics such as age, gender, and socio-economic status (Baumer, 2013; Spohn, 2000; Steffensmeier & Demuth, 2000; Steffensmeier et al., 1998; van Wingerden et al., 2016). Nonetheless, prior research has focused almost exclusively on the disadvantages of being Black or Latino defendants at multiple stages of the criminal justice processing, not on the advantages of being processed as other racial/ethnic groups such as Asians. To nurture the legitimacy perception among the general population, however, it is imperative to assess and improve the impartiality of sentencing decision in both aspects. That is, the discretion of prosecutors and judges should be exercised in ways that not just unfairly disadvantage some minority defendants, but advantage other minority groups. For example, if Black and Latino groups perceive that Asian defendants are sentenced more leniently on some unjustifiable grounds such as stereotyping, they are less likely to trust legal authorities and be motivated to comply and cooperate voluntarily with the legal authorities.
In the past decades, the proportion of Asian population in the U.S. has been growing steadily.
The recent U.S. Census data indicate that the estimated number of people who identified as Asian alone or in combination in the U.S. has increased from 17 to 24 million between 2010 and 2020, which account for 5.6 and 7.2 percent of the population, respectively (U.S. Census Bureau, 2020). However, relatively less is known about how Asian defendants are treated in the criminal justice system (Johnson & Betsinger, 2009; Kutateladze et al., 2014; Sampson & Lauristen, 1997; Spohn, 2000), largely because (1) most research (Albonetti, 1991; Steffensmeier et al., 1998) tried to explain why Blacks are overrepresented in multiple stages of criminal justice system and (2) there are substantially fewer Asian offenders than other racial/ethnic groups to warrant any valid and reliable estimation of Asian effects in the empirical analyses.
To investigate these relatively understudied issues, we analyzed a sample of felony offenders (n=11,626) indicted by the District Attorneys of the New York County between 2013–2017. The purpose of this study is two-fold. First, do we have evidence that judges are more lenient to Asian felony defendants when they make sentencing decisions possibly by stereotyping them positively as ‘model minority’? This study addresses this question by crafting improved measures of discretion and legal factors from a relatively recent and comprehensive dataset. Second, to what extent do earlier decisions made by prosecutors and judges affect the sentencing disparities observed at the final stages of sentencing decision? In particular, the current research attempts to fill the gap in the literature by incorporating bail request and pretrial detention decisions before indictment – critical ‘front end’ components of the adjudication process – as primary predictors of plea bargaining and imprisonment outcomes after indictment. This study contributes to extant literature in several meaningful ways. Prior research lacks comprehensive and precise measures of key legal factors, rendering the findings subject to omitted variable bias. Also, by focusing on the immediate offender and case characteristics, prior research overlooked the impact of the earlier decision makings that are less visible and even unavailable in the data, limiting our understanding of racial/ethnic disparity issues to the final sentencing outcomes.
Literature Review
In his minority threat thesis, Blalock (1967) proposed that as minority populations grow larger in size, dominant majority groups attempt to suppress them because they pose threats to the majority’s interests as well as existing social order and power arrangement more broadly (see also Chambliss & Seidman, 1982; Liska, 1992). Formal social control exercised by criminal sanctions is one of the major instruments in suppressing the threat of rising minority populations (e.g., Myers, 1990; Tittle & Curran, 1988). Initially, economic competition and power conflict were presumed to be the primary sources of threat associated with the increasing size of minority populations (Blalock, 1967). Accordingly, it is reasonable to presume that the fastest growing Asian populations in the U.S. should also constitute a rising threat to white majorities as black and Latino minorities did. In this vein, Asian defendants are expected to receive harsher criminal sanctions than white defendants. More recently, however, researchers have proposed an alternative causal mechanism linking the relative size of minority groups to disparate criminal justice outcomes. For example, Chiricos et al. (2006) hypothesized ‘fear of crime’ as a primary mediating factor in the minority threat hypothesis. In this vein, Asian defendants might be punished more leniently than their counterparts because they are perceived to be less threatening to the majority groups.
Focal concerns perspective is one of the most widely applied theoretical framework in the study of racial/ethnic disparities in the criminal case processing. This perspective contends that criminal justice actors consider primarily the harm caused by the crime, the defendant’s level of culpability, and perceptions of future dangerousness when they make decisions. In addition, practical concerns like organizational capacity to deal with heavy caseloads and the maintenance of working relationships among courtroom actors are presumed to influence decision making (Steffensemeier et al., 1998).
Drawing on the organizational theory of uncertainty avoidance/attributional decision-making (Albonetti, 1991; Hawkins, 1981), Steffensemeier and his colleagues proposed that, due to ‘bounded rationality’ resulting from insufficient time and information to make the most appropriate decision, courtroom actors tend to rely on ‘patterned responses’ by stereotyping offenders based primarily on offender’s blameworthiness, community protection, and other practical considerations (Steffensmeier & Demuth, 2000; Steffensmeier et al., 1998). To process a high volume of criminal cases efficiently under the inherent limitations of scarce resources and uncertainty in the prediction of future behavior, prosecutors and judges develop “perceptual shorthands” (Hawkins, 1981) that they can quickly apply after considering both legal and even extra-legal factors. As a result, Black and Latino defendants are perceived to be blamed than their white counterparts (e.g., engaged in more serious and violent types of offenses, having more aggravating and less mitigating factors as well as prior criminal history), posing a greater threat to the community (e.g., reoffending awaiting trial or after being released from incarceration), and thus have a high priority for severe punishments within the limitation of scarce resources (e.g., higher bail request, more pretrial detention, more upward guideline departure or less downward departure, less plead to a reduced charge, more imprisonment, more mandatory minimum/three strikes sentences, longer prison terms) (Johnson & DiPietro, 2012; Spohn & Holleran, 2006; Steffensmeier & Demuth, 2000, 2001).
Recent research adopted focal concerns perspective to explain why Asians are punished more leniently than Blacks and Hispanics, or even Whites at multiple decision-making points in the criminal justice system (e.g., Franklin & Fearn, 2010; Franklin & Henry, 2019; Johnson & Betsinger, 2009; Kutateladze et al., 2014, 2016; Lin et al., 2022; Wu, 2014, 2021; Wu & Kim, 2013). Unlike early Asian immigrants who arrived the U.S. at the turn of the 20th century looking for cheap labor and thus were perceived as Black slaves (Bonacich, 1984; Hirschman & Wong, 1986; Johnson & Betsinger, 2009), many affiuent, skillful, and well-educated Asians are recently coming to the U.S. looking for better education and job opportunities for themselves and their children (Hirschman & Wong, 1986). Because they perceive a greater sanction threat (Paternoster, 2010) than the general public due to their non-US citizenship and accrue more “stakes in conformity” (Toby, 1957) by working hard to materialize their American dreams – such as educational and financial success – through legitimate channels like higher levels of education and employment (Pew Research Center, 2013; Reeves & Bennett, 2004), they are more likely to stay out of trouble, comply with laws, and cooperate with legal authorities. Indeed, Asians are relatively underrepresented in the official crime statistics (Reeves & Bennett, 2004; Sampson & Lauritsen, 1997), and Asian adolescents are known to be less prone to deviance and delinquency than other racial/ethnic groups (Jang, 2002). Especially “Asian second-generation advantages” (Lee & Zhou, 2015; Tran et al., 2018, 2019) help them achieve better educational and occupational outcomes by accumulation of financial and social capitals, contributing further to the positive stereotypes associated with Asians during the criminal court processing.
Accordingly, Asians are generally perceived as “model minority” who are mostly hardworking and law-abiding citizens, and depicted in and reinforced by the media as such. In the criminal court, prosecutors and judges tend to associate Asian offenders more with non-violent misdemeanor or mere immigration-related offenses than with serious types of violent or drug offenses (Ho & Jackson, 2006; Hurh & Kim, 1989; Kasindorf & Chin, 1982; Lee, 1999; McGowan & Lindgren, 2006; Osajima, 1988; Paek & Shah, 2003; Wong et al., 1998). Due to the model minority stereotypes in which sufficient informal social controls are presumed to be working reasonably well, Asians are presumed to behave properly under sufficient and well-functioning informal social control mechanisms, and thus are less in need of formal social control than other racial/ethnic groups in the U.S. That is, Asians are less focal concerns to courtroom work group members because they are:
“seldom blamed for the serious crime problem in the United States, are predicted to pose a low level of danger to the community, and are perceived to have a low priority for incarceration in order to decelerate prison overcrowding (Franklin & Fearn, 2010; Johnson & Betsinger, 2009)” (Wu & Kim, 2013, p. 307).
More recently, Ulmer et al. (2022) in their critical examination of the focal concerns perspective from the perspective of judges found that judges consider more than those three concerns proposed initially by Steffensmeier et al. (1998) in their decision making. For example, judges also consider rehabilitation potential or redeemability very seriously as an end itself, not just as a means to better protect community by reducing recidivism. In this vein, Asian defendants might be less severely punished because judges view Asians have greater potential for rehabilitation due to the relatively higher sensitivity to sanction threats and higher stakes in conformity as well as well-functioning informal control mechanisms and support from their family and community. These perspectives are in line with the theories and models of decision making under bounded rationality frequently discussed in the public policy and administration literature (e.g., March, 1978, 1994; March & Simon, 1958; Simon, 1957, 1972, 1987).
Extant research has underscored the need to examine the effects of initial decision making on downstream outcomes within criminal justice processing (e.g., Baumer, 2013; Concannon & Na, 2023; Kurlychek & Johnson, 2019; Kutateladze et al., 2014). The focus on the last stages of the process discounts the influence of prior actions by courtroom actors and the previous outcomes that resulted in a final disposition and sentence. In this vein, to examine just a single point in the middle or end of the justice process might mask disparities that originate at other points. Increasingly, scholars are taking note of the successive points of disparate treatment for minority groups during the justice process (Concannon & Na, 2023; Kutateladze et al., 2014). To move the field forward, researchers have recommended analyzing prosecution as a “dynamic set of interrelated decision-making points” (Baumer, 2013). If the field has collectively found that racial and ethnic disparity exists at the final stages, it stands to reason that such disparities exist in earlier decision making and are worthy of further study. For example, the full impact of bail and pretrial detention is not just in their immediate effects. Defendants who cannot afford bail and thus are held in pretrial detention are more likely to plead guilty,1 and those who plead guilty are more likely to be convicted, incarcerated, and receive longer sentences after conviction (Sacks et al., 2015; Wooldredge, 2012). Thus, bail and pretrial detention acts as influential mediators that significantly influence downstream outcomes. Over the past decade, the number of studies examining the impact of upstream decision making has increased, but the number still pales in comparison to the sentencing-focused research.
More importantly, unwarranted disparate outcomes against minority groups are most likely to be produced at the early stages of criminal case processing because patterned responses such as stereotyping occur frequently when the prosecutors and judges are restricted to the highest level of ‘bounded rationality’ resulting from insufficient time and information to make the most appropriate decision. Tversky & Kahneman (1974) claim that many decisions are made based on the belief determined by heuristics such as representativeness when the probability of an event is uncertain. Relying heavily on a limited number of heuristic principles and rules, however, may lead to systematic biases and errors (p. 1124). Kahneman (2011) distinguished two modes of cognitive processing: System 1’ which involves fast and automatic thinking and decision relying primarily on intuition and heuristics shaped by prior patterns and experiences. ‘System 2’ is slow and deliberate in nature, requiring more thought and consideration than system 1 in the presence of sufficient time and information. Drawing on these discussions, it is reasonable to presume that upstreaming decisions are more prone to systematic biases because system 1 thinking and decision would dominate at the initial stages of case processing than during the final phases.
In a similar vein, in her more recent critique of the focal concerns perspective, Lynch (2019, pp. 1155-1156) argues that there is less room for these focal concerns to affect judge’s exercise of discretion at the formal sentencing stage of the criminal justice process because judges are acting in more public settings and have more information about defendants and time to review materials rendering stereotyping less necessary. Most of all, sentencing decisions at the later stages of criminal processing are not totally under judge’s control because sentences tend to be standardized constrained by laws, going rates, and binding pleas limiting judge’s discretion. Thus, attribution and stereotyping practices, if they exist, are more likely to happen at earlier stages which are less regulated and more hidden (Lynch, 2019).
Despite the growing Asian population in the U.S., the vast majority of attention has been afforded to the comparison between similarly situated Blacks, Latinos, and White offenders. Although most research finds Black and Latino defendants are punished more severely than their white counterparts net of legal factors, little is known about how Asian offenders are being processed in the criminal justice system (Johnson & Betsinger, 2009). Recently, a group of scholars (e.g., Everett & Wojtkiewicz, 2002; Franklin & Fearn, 2010; Franklin & Henry, 2019; Johnson & Betsinger, 2009; Rodriguez, 2003) began to expand the scope of the research and theorizing on the relative treatment of Asian offenders to further our understanding of how race and ethnicity affect discretionary outcomes in the judicial decision making.
Everett & Wojtkiewicz (2002) in their analysis of federal sentencing process found that, although Blacks, Hispanics, and Native Americans are subject to harsher sentencing outcomes than similarly situated Whites, Asians and Whites were treated almost equally. However, this study focused only on the final sentence length decision, leaving many other critical questions unanswered – such as how earlier decisions affect subsequent outcomes. Similarly, Rodriguez (2003) also found that Asians were treated no differently than similarly situated Whites after assessing the effect of prior strikes on the sentences of felony offenders under Washington State’s guidelines.
Contrary to these early findings that Asians are treated similarly as Whites at the final stages of sentencing decision especially when sentencing guidelines are in place, Johnson & Betsinger (2009) in their comprehensive analysis of Asian offenders in the federal courts found that Asian offenders were more likely to receive substantial assistance departures, but less likely to be incarcerated than Black, Hispanic, and even White offenders. Similarly, Franklin & Fearn (2010) found that Asian offenders are treated more leniently in the imprisonment decision after extending this line of research at the state level. Kutateladze et al. (2014) also found that Asian defendants are the least likely to be detained awaiting trial, to receive custodial offers, and to be imprisoned among different racial and ethnic groups in the New York County. Using a unique sentencing outcome other than imprisonment and sentence length, Franklin & Henry (2019) also found that Asian offenders are less likely to be denied access to “good time” through sentencing.
Despite the contribution made by these seminal studies, the empirical investigation on the relative sentencing of Asian offenders is still in its infancy. In particular, there is a paucity of research on how Asians are sentenced in state courts because most prior research has focused on the sentencing of Asian offenders in federal courts. Although focal concerns can vary substantially across different socio-legal contexts (Johnson, 2005, 2006; Lynch, 2019), there might be meaningful variation at the micro level within the same jurisdiction sharing similar characteristics. Additionally, most research did not explicitly account for the significant and lingering effects of earlier decisions on the final sentencing disparity outcomes. In light of this unique status, an analysis of Asian defendant’s experiences with criminal justice processing is warranted and sorely needed. Drawing on the recent research findings (Franklin & Fearn, 2010; Franklin & Henry, 2019; Johnson & Betsinger, 2009; Kutateladze et al., 2014), we hypothesize that Asian defendants would be subject to less punitive outcomes than White defendants in New York County even after controlling for the confounding effects of relevant covariates and early decision-making outcomes.
The theoretical foundation and empirical base of this research suggest a consistent pattern of advantage for Asian defendants in multiple phases of criminal case processing and give rise to the following hypotheses.
-
Asian defendants would be subjected to more advantages of receiving favorable plea bargains than similarly situated White defendants.
-
Asian defendants would be subjected to more advantages of receiving less sentences to incarceration than similarly situated White defendants.
-
Earlier decision outcomes are significant predictors of downstream outcomes.
Data and Method
The current dataset is unique in terms of its recency, comprehensiveness, locality, and validity of key variables. The dataset includes nearly all felony cases recorded in the New York County District Attorney’s Office (DANY)’ case management system between 2013 and 2017 and was prosecuted by one of six ‘Trial Bureaus’ at DANY. The DANY’s case management system was established almost 30 years ago by its own Information Technology department and has been extensively modified to address the shortcomings and accommodate newly emerging demands. Although the system contains rich information about the offender and other case characteristics and outcome measures of the case processing, we supplemented it with more comprehensive and precise criminal history data obtained from the New York State Department of Criminal Justice Services containing all New York State convictions. We analyzed a sample of 14,601 defendants who were indicted during 2013–2017.
Thus, the current dataset allows for a rigorous examination on the effect of race and ethnicity on the final sentencing outcomes by enabling us to control for valid and reliable measures of legal and extralegal factors as well as previous case processing outcomes. In particular, the data extracted for this study contained measures of five discretionary decision points marked sequentially over the life of a case. Using the initial and intermediate case processing outcomes as key predictors allows for a more complete understanding of the sources of disparate outcomes at the final stages.
The dependent variables are favorable plea bargaining and incarceration outcomes. Favorable plea Bargaining is operationalized as a reduced charge (coded as 1) compared to the original indictment charge (coded as 0) applied as a result of plea bargain. Prosecutors have significant discretion regarding the charges to which they allow the defendant to plead. Prosecutors may insist the defendant plead to the ‘top charge’ (the most serious charge on the complaint) or may offer a reduced plea, depending on the level of harm and culpability of the defendant. Plea bargaining is important because most cases are resolved via guilty pleas (Johnson, 2003; Kutateladze et al., 2016; Smith, 1986) but seriously understudied due to its less visible nature. Incarceration is coded as 1 when a defendant received a sentence of jail or prison, and 0 when a defendant was sentenced to anything else. However, prosecutor’s sentencing recommendations are not captured in this data and could not be incorporated into the analytic models.
The primary independent variables are the race or ethnicity of the indicted defendant. We created multiple dummy variables for White, Black, Latino, and Asian defendants and used White as the reference category. Other key demographic characteristics of the defendants were also included as control variables. Gender is a dichotomous variable coded as 1 for male defendants. Age is measured at the ordinal level with the categories of under 18, 18–24, 25–35, and above 35 (25–35 serving as the reference category).
Numerous additional independent variables were included to control for the effects of legal factors such as the severity of the offense, case characteristics, and the defendant’s criminal history. Instead of relying on the simplistic measures widely adopted in prior research, we improved these measures by collecting additional and more detailed information to better capture the meaningful variations in these legitimate factors that prosecutors and judges consider most significantly in their decision making. Then, we created multiple dummies after consulting with practitioners who confirmed that these categories are meaningful and reflect how they consider them in practice.
For the types and severity of the offense, we constructed multiple dummy variables categorized by the combination of the statutory severity, presence of violence, and offense type for the top 13 most common charges: Class B drugs, Class D violent assault, Class D forgery, Class C violent robbery, Class D grand larceny, Class D burglary, Class D drugs, Class D weapons, Class D violent Burglary, Class E contempt, Class B violent robbery, Class E theft, Class D robbery. All other charges were grouped into an ‘Other’ category. After conferring with stakeholders, we chose Class E grand larceny offense as the reference category in the analysis. We also measured and controlled for other defendant and case characteristics such as whether the defendant had another pending case in Manhattan at the time of the instant offense, whether the case was flagged as domestic violence, and the defendant’s defense attorney type, all of which are also known to affect the outcomes.
The criminal history of the defendant was measured by four distinct categories: prior felony, misdemeanor, violent convictions, and the number of bench warrants issued for the defendant. Further, these counts were collapsed or ‘binned’ into groups that prosecutors identified as meaningful. Absence of these types of criminal history served as the reference category in the analysis. These criminal history measures are also significantly different from those adopted by prior research, especially from the same jurisdiction. For example, Kutateladze et al. (2014) included measures of prior arrests and prior incarcerations, and these records were only for prior Manhattan arrests and incarcerations because statewide data were not available at the time of their research. The data for the current study include criminal convictions from throughout New York State, provided by the New York State Department of Criminal Justice Services. Another set of defendant characteristics was whether the defendant was tracked internally by the Crime Strategies Unit as a ‘crime driver’ in Manhattan. A specialized software known as the ‘Arrest Alert System’ helped prosecutors and analysts organize persons of interest and push out notifications if any of those individuals are arrested in New York City (Tallon et al., 2016). Relatedly, the Arrest Alert System also contained information about whether the defendant was under New York State parole supervision at the time of the offense.
As discussed in the previous sections, this study also controlled for the effects of the prior decision-making outcomes before indictment such as prosecutor’s bail request and judge’s pre-trial detention decision. Bail request is the bail amount requested by the assistant district attorney during initial case screening and prior to criminal court arraignment. Pre-trial detention is coded as 1 for defendants who had some amount of monetary bail set at criminal court arraignment and coded as 0 for defendants who were released on their own recognizance.
To investigate racial and ethnic disparity in the court decision-makings, we adopted a set of multivariate logistic regression models for each of the two discretionary outcomes. Logistic regression models were used because both dependent variables were binary outcomes. Per each outcome, we created four regression models to take a stepwise approach: the first with only race and ethnicity as an independent variable, the second with race/ethnicity and all legally relevant and other important control variables, the third with race/ethnicity, control variables, and measures of prior decision-making, and fourth with Heckman’s correction for selection bias.
Considering that the current data were restricted only to a ‘indicted’ subset of felony cases, selection bias might exist when the indicted cases are substantially different from the not-indicted cases. We addressed this issue by using Heckman’s correction for selection bias (Heckman, 1979), which accounts for potential biases introduced when only indicted offenders are analyzed out of the entire offender sample. This approach is known to be useful when examining racial/ethnic disparity at various stages of criminal justice processing (Wooldredge & Thistlethwaite, 2006). These adjustments might improve the performance of the regression models by estimating more precise estimates of racial and ethnic coefficients as well as the unique effect of prior decision-making on the final sentencing outcomes.
Results
Table 1 reports descriptive statistics for the dependent and independent variables summarized by race and ethnicity. Approximately eighty-five percent of the defendants in the current sample were male and the average age was of 33. On average, defendants had just under one prior felony conviction and five prior misdemeanor convictions. In addition, they had failed to appear approximately 1.8 times on average in a prior New York County prosecution. According to available criminal history data, eighty-three percent of defendants had no prior violent convictions, roughly seven percent were under New York State Parole supervision, and twelve percent were flagged in internal intelligence databases as gang members or suspected ‘crime drivers.’
In addition to the defendant and offense characteristics, prior discretionary decisions made by prosecutors and judges before indictment were also measured as key independent variables. The mean and median values of bail request were just under $24,000 and about $15,000, respectively, but there were some notable differences across four racial and ethnic groups. Asian and White defendants had a median bail request of about $10,000, compared to $15,000 for Black and Latino defendants. A log transformation was applied to the bail request variable due to its significant skewness. Approximately 70 percent of defendants were detained awaiting trial after arraignment, with Black defendants (72%) most likely to be detained, and Asian defendants (50%) least likely to be detained. After criminal court arraignment, just over 40 percent of defendants were indicted, about 30 percent received a favorable plea bargain, and 67 percent were imprisoned. Consistent with the patterns observed in the prior research, Asian defendants were at the lowest risk of receiving punitive outcomes. Only 29 percent of Asian defendants were indicted, 36 percent received a favorable plea bargain, and 39 percent were imprisoned. The earlier disparate outcomes might at least partially account for the racial/ethnic disparity observed in the final sentencing outcomes of our primary interest.
Although the descriptive statistics suggest significant variation in the dependent variables between race and ethnicity categories, any disparities might result from the differences in legally relevant indicators of offense severity, offender’s criminal history, or other extra-legal indicators such as the type of defense attorney. To account for these pre-existing conditions and thus identify the unwarranted racial/ethnic disparities more precisely, we next estimate a series of multivariate logistic regression models with varying sets of covariates: Model 1 includes only race and ethnicity, Model 2 adds legally relevant and other extra-legal variables to Model 1, Model 3 adds measures of upstream decision making outcomes to Model 2, and Model 4 applies Heckman’s correction to Model 3 to further account for any selection bias that might exist when analyzing only indicted cases out of the entire sample. Model diagnostic tests indicate no harmful levels of multicollinearity, with the variance inflation factors among all the predictors being below the standard ceiling of four (Wooldridge, 2009).
During the final adjudication process after indictment, cases are disposed of via a plea of guilty or a conviction because the policies of the Manhattan District Attorney’s Office do not allow a felony defendant to plead guilty to a felony charge without being indicted. We first estimated the likelihood of pleading to a lesser or reduced charge compared to the ‘top’ charge on the indictment and compared the odds of favorable plea between different racial and ethnic groups after relevant controls. Then, we compared the odds of imprisonment after fully accounting for the effects of legal and extra-legal factors, as well as prior discretionary outcomes including plea bargaining.
Table 2 reports the odds ratios and standard errors estimated from varying model specifications for both outcome variables. The first model – with only race and ethnicity to predict the likelihood of a reduced plea – suggests that Asian defendants are associated with a substantially and significantly increased odds of a lesser plea compared to similarly situated White defendants (odds ratio=1.58, p<0.001). Unlike the Black and Latino effects which are not statistically significant after adding control variables in Models 2–4, the pattern of favorable plea outcomes for Asian defendants persists even after controlling for the same legal and extra-legal variables (Model 2) and measures of prior decision-making outcomes (Model 3), as well as Heckman’s correction for selection bias (Model 4).
The next set of models compared the odds of convicted defendants being sentenced to new terms of jail or imprisonment (vs. non-incarcerative sentence). The race/ethnicity only model (Model 1) shows that Black and Latino defendants are more likely to be imprisoned whereas Asian defendants are less likely to be imprisoned than White defendants (odds ratio=0.47, p<0.001). Such disparate outcomes remain statistically significant even after introducing relevant control variables, prior decision-making outcomes, and Heckman’s correction in Models 2–4, although the Asian effects become less obvious in more sophisticated model specifications. After the addition of legal and extra-legal variables (Model 2) and prior decision-making outcomes (Model 3), neither Black nor Latino defendants were associated with a statistically significant difference in the odds of imprisonment. However, Asian defendants were associated with decrease in the odds of incarceration compared to White defendants by 30 percent in Model 2 (odds ratio=0.70, p<0.001), 34 percent in Model 3 (odds ratio=0.66, p<0.01), and 9 percent in Model 4 (odds ratio=0.91, p<0.01). Among control variables, characteristics that increased the likelihood of incarceration followed a similar theme as plea bargaining as an outcome models. Prior misdemeanor and felony convictions and parole or ‘crime driver’ status significantly increased the odds of incarceration. Defendants charged with all types of violent offenses were significantly more likely to be sentenced to jail or prison. The factors that made defendants less likely to be incarcerated were drug, forgery, and vehicle crimes, as well as domestic violence cases and defendants under 18 or above 35 years old.
In addition to the racial/ethnic effects, the outcomes for three measures of prior decision making – bail request, pretrial detention, and whether the defendant pled to a lesser charge – are noteworthy as discussed in the previous sections. As expected from prior research (Concannon & Na, 2023; Kutateladze et al., 2014), defendants with increased bail requests (odds ratio=1.75, p<0.001) or who were held in pretrial detention (odds ratio=3.12, p<0.001) were associated with a higher risk of incarceration. Model 4 applied the Heckman correction and produced similar results. Although all the odds ratios were subdued in magnitude, they remained statistically significant. Defendants who pled guilty to a reduced charge were significantly less likely to be sentenced to incarceration. Model 3 estimated defendants who received a plea bargain were 17 percent less likely to be sentenced to jail or prison (odds ratio=.83, p<0.001), while the Model 4 estimated these defendants were just two percent less likely to be imprisoned (odds ratio=.98, p<0.05). This is understandable given the consistent finding that negotiated pleas produce significant discounts in sentence severity (Piehl & Bushway, 2007). Further, the research community has identified prosecutorial discretion as a significant predictor of sentencing outcomes, and that judges often rely on prosecutor’s sentencing recommendations when considering the final sentence (Devers, 2011; Kurlychek & Johnson, 2019). It stands to reason that defendants who were granted a plea bargain by the prosecutor also negotiated some aspect of the sentencing recommendations, and the plea bargain had a significant impact on the imprisonment outcome (e.g., binding pleas: Lynch, 2019).
Discussion and Conclusion
The current study investigated racial and ethnic disparity observed after indictment using a unique and robust dataset of nearly all felony cases commenced between 2013 and 2017 in New York County (Manhattan), New York. Drawing on focal concerns perspective and prior research, we hypothesized that Asian defendants would be treated less harshly than similarly situated White defendants, and early decision-making outcomes would be important sources of such disparity. The results of our multivariate analyses provided strong evidence supporting these hypotheses. Consistent with our prediction, Asian defendants received significantly more favorable plea bargains and less imprisonment sentences even after accounting for a set of legal and extra-legal factors. In addition, similar to the patterns observed in recent research on cumulative disadvantage (e.g., Kurlychek & Johnson, 2019; Kutateladze et al., 2014; Stolzenberg et al., 2004), prior decision-making outcomes such as bail request, pretrial detention, and whether the defendant pled to a lesser charge had significant impacts on the final imprisonment decision.
Our findings lend credence to the continued debate on the differential effects of race and ethnicity on various prosecutorial and judicial outcomes. They also reaffirm that many discretionary decisions made during the early phases of the adjudication process are important own their own, but also have lasting effects on the subsequent decisions. Along with race and ethnicity, legal indicators such as crime severity and prior criminal history were strongly associated with more punitive outcomes, findings that are also consistent with a vast majority of prior research (e.g., Stolzenberg et al., 2004; Sutton, 2013). These findings suggest that, under the inherent limitations of insufficient time and information, prosecutors and judges use a combination of not just legally relevant indicators of offense severity and defendant’s history, but also other extra-legal defendant characteristics in their decision making (Albonetti, 1991; See also Lynch, 2019; Steffensmeier et al., 1998; Ulmer et al., 2022).
Theoretically, our findings are consistent with the propositions of uncertainty avoidance (Albonetti, 1991) and focal concerns (Steffensmeier et al., 1998; Ulmer et al., 2022) perspectives. Asian defendants are less likely to experience punitive outcomes, as the group as a whole has not been tied to negative criminal stereotypes like Black and Latino individuals (Steffensmeier et al., 1998). Rather, they may be subject to positive stereotypes because judges perceive Asian defendants have a greater chance for successful rehabilitation due to their strong informal social control and support mechanisms such as family support and community ties (Ulmer et al., 2022). Similar findings in the prior research also have been theorized to be the result of a translation of positive stereotypes of Asians into positive outcomes for Asian defendants, and the opposite effect for Black and Latino defendants (Johnson & Betsinger, 2009; Kurlychek & Johnson, 2019; Kutateladze et al., 2014).
The sentencing disparities observed in plea bargain and imprisonment might also reflect the accumulation of small and often invisible advantages or disadvantages in a series of decision makings by prosecutors and judges that precede indictment (e.g., Concannon & Na, 2023; Kutateladze & Andiloro, 2014; Kutateladze et al., 2014). Many of those early decisions occur promptly based on relatively insufficient information about the defendants and case characteristics. More seriously, such expansive prosecutorial and judicial discretions are subject to less formal rules and regulations, being exercised largely out of public view and media oversight. In this vein, focusing exclusively on the highly visible, final sentencing outcomes might mask a more complex, subtle, and nuanced process of compromising racial and ethnic neutrality during the entire adjudication process. Even after a series of efforts to remedy unwarranted disparities in sentencing outcomes (e.g., introducing mandatory or presumptive sentencing guidelines), racial/ethnic minorities are still overrepresented at all stages of adjudication – with some notable exception of Asians. It might result from the fact that discretion can be shifted to the earlier stages of the adjudication process (Alschuler, 1978; Miethe, 1987; Nagel & Schulhofer, 1992) who still resort to ‘patterned responses’ based not just on legitimate legal factors but also other extra-legal factors such as race and ethnicity of the defendants – even after a series of recent legal reforms that bind courtroom actors to follow formal rules and regulations such as sentencing guidelines with two-dimensional grid or matrix (using offense severity and prior criminal history) or established organizational practices to reduce unwarranted sentencing disparity (e.g., Ashworth, 2015; Frase, 2005; Lofstrom et al., 2020). Thus, implicit biases might still emerge without conscious awareness of the courtroom actors who are required to make numerous decisions under inherent constraints of insufficient time and information. Compared with final sentencing decisions, these early processes involve more discretionary practices under little oversight from internal supervisors or the public. To address the weakness of prior research that has limited its spectrum to a single decision-making point, the current study considered earlier decision-making outcomes simultaneously for more complete understanding of the sources of disparities in the sentencing outcomes.
The current study also provides insight into contemporary racial and ethnic disparity issues in criminal justice. The results of this study are directly applicable to conversations about racial equity, bail reform, abolition of pretrial detention for less serious offenders, and related criminal justice policy debates. This research uncovered that the requests prior to the bail decision and pretrial detention decision had significant impacts on the final sentencing outcomes. These findings add nuance to a body of literature that consistently finds monetary bail schemes disproportionately affect specific racial and ethnic groups. This research also adds to the body of literature by finding that defendants detained awaiting trial are less likely to receive favorable plea bargains, and more likely to be sentenced to terms of incarceration. Bail reform is a broad topic and efforts to reduce disparity are found in many forms, such as diversion programs, risk assessments, alternatives to detention. Acknowledging and remediating these disparities through bail reform or other mechanisms is the first step towards creating a fair and equal criminal justice system. Considering that unfair criminal justice system involves not just unduly harsh punishments but also lenient punishments for specific segments of population, reform efforts need to be followed when researchers continue to discover the evidence of unwarranted racial disparities in the criminal case processing outcomes.
Despite significant contributions, this study also has some limitations. First, although sentencing outcomes are known to vary not just at the individual case level but also at the judge, courtroom, and the community levels (Johnson, 2005, 2006; Lynch, 2019), this study only analyzed the case level characteristics assuming that these cases are independent with each other. Future research should assess contextual disparities using multilevel analyses. Second, considering that Asian Americans are composed of diverse ethnic groups with varying socioeconomic status as well as differing educational and career outcomes, the overall Asian effects estimated in the analytic model might mask substantial variations that might exist across distinct Asian subgroups. Future research should explore whether Asians of different nationalities or different immigration status (e.g., U.S. citizens: Wu & Kim, 2013; Wu, 2021) are receiving more or less favorable treatment at multiple stages of criminal justice system to verify the validity and applicability of focal concerns perspective in the study of racial/ethnic disparity (Johnson & Betsinger, 2009). Future research should also assess whether the offense type or people’s perception of (or actual) immigration status of Asian offenders affects the core proposition of the model minority thesis (Wu & Kim, 2013). Most of all, for more comprehensive assessment of the validity of focal concerns perspective and its applicability to Asian defendants, future research should directly measure judge’s cognitive decision-making process and investigate what judges really consider is linked to the observed demographic disparities in the sentencing outcomes instead of merely speculating on it (Lynch, 2019, p. 1159; see also Ulmer et al., 2022). Finally, despite our efforts to improve the quality and quantity of control variables, there still might be other confounding factors that might explain the observed disparities, such as socioeconomic status or specific cultural factors within Asian communities. Future research should incorporate more direct and valid measures of these factors to better disentangle the sources of disparate patterns in sentencing outcomes.
