Criminal Justice System Reforms to Reduce Racial Inequality
Criminal justice policy has been an active area of reform over the last two decades. Proposals to reduce criminal justice disparities often focus on the specific stages of criminal processing. The committee views reforms to police, courts, and corrections as part of a larger transformation needed to significantly reduce racial inequality. Reducing arrests, incarceration, and other criminal justice contact in Black, Latino, and Native American communities will require reforms not only in criminal justice policy but also in other policy domains.
Research discussed in Part I shows that criminal justice contact, from police stops through arrests, prosecution, and incarceration, was disproportionately focused on predominately Black and Latino communities throughout the 20th century and became pervasive in the 1990s and 2000s. Arrest and incarceration became common responses to social problems related to concentrated poverty in communities marginalized by segregation. Chief among these problems is serious violence, which has taken a vast toll on disadvantaged communities, which are disproportionately Black, Latino, and Native American. But drug use, untreated mental and physical illness, failing schools, housing insecurity, and cumulative disadvantage over the life course and across generations have also created pathways into the criminal justice system. Part I also provided evidence that punitive policy disproportionately harmed Black, Latino, and Native American communities. Stops and searches and the use of force, sometimes fatally, by police combined with the communitywide effects of high rates of incarceration have transformed the relationship between state and citizen and undermined trust in authorities.
Reducing racial inequality in the criminal justice system holds the promise of improving safety by reducing the harms of interpersonal violence as well as the harms of policing and incarceration. Promoting racial equity will partly depend on abandoning a conception of community safety that relies mainly on punishment. In recent years, many communities and even criminal justice agencies have begun to move away from such a model, sometimes adopting alternative interventions where the threat of punishment becomes just one of several possibilities. As an example, specialty courts throughout the nation focus on the needs of people with severe mental illness, substance abuse problems, and veteran status, and others who are highly involved in the criminal justice system and have identifiable unmet needs. In another example, as discussed in Chapter 6, anti-violence strategies that rely on violence interrupters and other community members devoted to deescalating conflicts have been functioning largely independent of local law enforcement.
This chapter describes the implications of the committee’s findings specifically for criminal justice system reforms. It reviews research on alternative interventions within criminal justice settings aimed at reducing violence and criminal justice contact, particularly in Black, Latino, and Native American communities. It pays close attention to recent changes in policy that have led to declines in police stops and correctional supervision. To be sure, eliminating racial and ethnic disparities in criminal justice involvement will require eliminating inequality in other domains. Hence, the criminal justice reforms discussed in this chapter should be viewed as one piece of a coordinated set of reforms that also extend beyond the criminal justice system.
Criminal justice agencies are charged with responding to and preventing serious interpersonal harm in communities. Traditional criminal justice policy in the United States pursues these objectives through retributive sentencing and a host of policies designed to either deter criminal activity or incapacitate the criminally active. Responding to harm runs through the history of policing and penal codes and is foundational to the jurisprudence of punishment. For its part, the process of punishment, decided by the courts and executed by correctional authorities, is treated as a form of accountability. Retributive theories of punishment contend that the pain of punishment proportionately imposed in the name of the people by the state restores moral balance after one person has harmed another.
Preventing crime is complex and difficult to assess, given the many determinants of criminal activity and the difficulty of specifying counterfactual levels of crime in the absence of a given policy. Social science research has focused largely on various strategies designed to either deter criminal activity or incapacitate individuals who would otherwise commit a crime. Deterrence strategies can be broad-based, such as through
the imposition of severe sentences, or quite focused, such as through hot spot patrols or swift-and-certain sanctions. Incapacitation strategies rely on surveillance and the suppression of liberty, with intervention ranging from regular check-ins with a probation officer to solitary confinement. Empirical researchers have tried to estimate the deterrence and incapacitation effects of incarceration and other forms of punishment (e.g., Nagin, 2013; Paternoster, 2010; Hawken and Kleiman, 2009).
Our starting point for describing criminal justice reforms that can reduce racial inequality is the proposition that the criminal justice system should minimize the overall harms from crime, including harms that result from society’s responses to crime. Minimizing the harms from crime expands the tool kit beyond retribution, deterrence, and incapacitation to include victim restoration, prevention through improved community relations, attention to unmet needs, and cross-system coordination with non-criminal justice agencies. Minimizing the harms caused by criminal justice officials would rule out indiscriminate, widespread, and unconstitutional stop-question-and-frisk practices in minority neighborhoods, frequent querying of citizens about their probation/parole status during minor traffic stops, and pretextual traffic stops for minor equipment violations, all practices that show little crime-reducing impact but which undermine community trust. Moreover, minimizing both the harms of crime and system harms would preclude long sentences that cannot be justified by incapacitation or deterrence. On the other hand, minimizing criminal and system harm may still yield a significant role for police in clearing homicide cases, reducing gun violence, and reducing the incidence of robbery, burglary, and other serious crimes that are disproportionately experienced in disadvantaged communities.
Given the large presence of the criminal justice system in minority communities, reducing both crime and system harms would have a disproportionate positive effect on impact, and likely reduce disparities in system involvement and victimization if implemented across the board or concentrated in Black, Latino, and Native American communities. However, if implementation is solely left to the discretion of criminal justice actors, it is possible to reduce average harm while not disproportionately reducing harm for affected racial and ethnic groups. One approach to reducing racial inequality has involved shrinking the scale of punitive impact of criminal legal processing—say, the numbers of arrests or prison admissions—in Black, Latino, and Native American communities. Shrinking the scale of punitive impact is different from many reforms that aim to reduce differential treatment, for example through anti-bias training.
We discuss criminal justice reform by first examining the structure of criminal processing and large-scale policy efforts focused on drugs and violence through the coordinated work of police, court actors, and the penal
system. We then consider each of the main stages of criminal processing—by police, in the courts, and in sentencing and corrections—and explore how they might be reformed to reduce the scale of punitive impact and thereby reduce racial inequality in a way that serves the mission of responding to community violence and other harm.
INSTITUTIONAL STRUCTURE AND POLICY
Before examining each of the main stages of criminal processing, it is useful to consider the overall structure of interlocking agencies that comprise the criminal justice system and how they might operate together in a coordinated way to reduce racial inequality.
Part I of this report provides evidence of large racial disparities in involvement with the criminal justice system at all stages. Racial disparity preceded the increases in incarceration that occurred during the latter half of the 20th century and is certainly still with us today. In fact, policies that increased the punitiveness of sentencing have aggravated absolute racial disparities, while policies that reduced the severity of sanctions have narrowed them. Hence, one strategy for reducing disparities would be to reduce the overall punitiveness of sanctions and to exercise greater parsimony and proportionality.
Part I also provided evidence that, net of the general punitive level of the criminal justice system, racial and ethnic minorities, Black Americans in particular, experience greater scrutiny by and more frequent, intrusive, violent, and lethal interactions with police. Moreover, analysis of police stops finds that whereas Black Americans are the most likely to be stopped by the police, they are often the least likely to be ticketed or arrested, suggesting that many stops are effectively pretextual. Black Americans are also more likely to be searched yet in many jurisdictions are the least likely to be discovered to be carrying contraband. These disparities likely reflect both differential treatment of Black citizens by specific officers as well as more intrusive policing in cities with large minority populations. A second strategy for reducing racial inequality thus involves addressing bias by criminal justice officials and identifying seemingly race-neutral policies with racially disparate effects.
Constitutional Sources of Parsimony
In a discussion of sentencing policy, the legal scholar Morris (1974) outlined a theory of parsimony that held that the penal power of the state should be used minimally, just sufficient to achieve the goals of public policy. Excessive punishment, he wrote, amounted to state cruelty. Later writing expanded the idea of parsimony beyond sentencing and imprisonment to
include policing, pretrial detention, and the manner of supervision on probation and parole (Atkinson and Travis, 2021).
The idea of parsimony—the minimal use of the state’s power of coercion—although regularly overlooked in practice, has constitutional roots in the Bill of Rights and later amendments that embody a philosophy of limited government, especially in the area of criminal justice. Thus, policing power is specifically regulated through the Fourth Amendment, which requires police to justify intervention on the basis of probable cause or reasonable suspicion and establishes protections against unreasonable search and seizure (e.g., Terry v. Ohio;1United States v. Cortez2). Sixth Amendment rights to a speedy trial, to trial by jury, and to counsel also put limits on overreaching prosecution and provide the constitutional basis for indigent defense (e.g., United States v. Wheat).3 The Eighth Amendment prohibits cruel and unusual punishment and has been applied to the conditions of incarceration, most notably to conditions of prison overcrowding and access to health care (e.g., Hutto v. Finney,4Brown v. Plata5). The Fourteenth Amendment’s equal protection clause prohibits racially discriminatory laws and policies and has been applied by federal courts to limit racial profiling by police (e.g., Floyd et al. v. City of New York et al.6).
Throughout this chapter we will see two main strategies for reducing racial inequality in the criminal justice system. One strategy reduces the scope of criminal justice contact by reducing the number of police stops, arrests, court appearances, and incarcerations. The second strategy reduces the disparate impact of policy on and differential treatment of people of color. Although discrimination and disparate impact have been a focus for many reform efforts to reduce racial inequality, we also find strong evidence of reductions in absolute inequality by reducing the scale, duration, and intensity of criminal justice intervention. Proposals for reducing the overall extent of criminal justice contact are consistent with earlier research-based recommendations for reducing incarceration (NRC, 2014) and also align with founding commitments to constitutional limits on state power.
Assessing Decisions and Costs across Levels of Government
Like all public organizations, local, state, and federal criminal justice agencies operate subject to budget constraints and are often responsive to cost incentives. Moreover, police agencies are often called upon to enforce
1 392 U.S. 1 (1968). See https://supreme.justia.com/cases/federal/us/392/1/
2 449 U.S. 411 (1981). See https://supreme.justia.com/cases/federal/us/449/411/
3 486 U.S. 153 (1988). See https://supreme.justia.com/cases/federal/us/486/153/
4 437 U.S. 678 (1978). See https://supreme.justia.com/cases/federal/us/437/678/
5 563 U.S. 493 (2011). See https://supreme.justia.com/cases/federal/us/563/493/
6 959 F. Supp. 2d 540 (S.D.N.Y. 2013). See https://casetext.com/case/floyd-v-city-of-ny-2
regulations that often generate revenue for local and state government through fines and fees. To be sure, a fine or fee is often less severe than incarceration or community supervision, and fines and fees may have a greater role in a less punitive system. However, if generating fine and fee revenue motivates enforcement, these sanctions may distort police behavior, directing policing toward those activities that generate revenue.
Fine and fee revenue provides an example of how financial considerations create incentives for excessive enforcement or harsh sanctions. When police departments share in the revenues from enforcement actions, the incentives are quite clear. There are many instances, however, where the structure of public financing provides less obvious incentives. For example, a probation department that relies heavily on supervision fees will have incentives to surveil compliant individuals on probation who are current on their fees for as long as possible. Localities that pay none of the cost of sending someone to prison yet bear the full cost of a local sanction have an incentive to foist punishment costs onto the state, and perhaps to overuse state prisons.
A notable example of revenue generation distorting enforcement activity was provided by the Department of Justice’s report on policing in the municipality of Ferguson (MO). Following the killing of a Ferguson resident, Michael Brown, by a local police officer in 2014 the Civil Rights Division of the Justice Department investigated the killing and the broader context of policing in the town. Like many U.S. towns (Graham and Makowsky, 2021), the local budget of Ferguson relied on municipal fines and fees (U.S. Department of Justice, 2015). Ferguson raised revenue with police-issued citations, and such efforts were focused on Black residents. Ferguson’s “revenue-driven policing” increased contact between police and the public, and police stops often carried the potential of escalation. The proliferation of criminal justice fines and fees has been found to expand contact between the police and courts on the one hand and between police and the public on the other in a large number of jurisdictions (Shannon et al., 2020). The accumulation of criminal justice debt triggers warrants for non-payments, court appearances, and third-party debt collection, which can prolong court involvement well beyond the term of the initial sentence (Pager et al., 2022; Pattillo and Kirk, 2021). In Ferguson, the Justice Department found that pressures from city officials to raise revenue through citations led to an unconstitutional style of policing that demanded compliance in the absence of legal authority, led to stops without reasonable suspicion, and led to arrests without probable cause (U.S. Department of Justice, 2015, pp. 28–68). These examples of probation fees and municipal citations suggest that reducing the dependence of public financing on criminal justice contact could reduce the public’s contact with police and the courts. Research has demonstrated that the political economy of law enforcement—the need to
raise revenue through the criminal justice system—exacerbates racial bias and the expropriation of wealth from politically vulnerable subpopulations through the criminal justice system (Goldstein et al., 2020; Makowsky et al., 2019; Sobol, 2017; Makowsky and Stratmann, 2009; Blumenson and Nilsen, 1998).
Reducing criminal justice contact by altering cost incentives between different levels of government has been demonstrated in California with reforms to the juvenile justice system in the 1990s and to adult sentencing in the 2010s. Besides reducing correctional populations with no measurable change in crime rates, the reforms in California show that counties can respond to both fiscal savings and costs. Juvenile justice reform illustrates the effects of changes to local costs. In 1996, the state legislature passed a bill that increased the monthly costs for juvenile admissions to the California Youth Authority, the state agency that at the time ran state juvenile corrections facilities. Prior to the legislation, counties paid $25 per month for each of the agency’s wards. Starting in 1997, the monthly payment increased to $150 per month for youth convicted of the most serious offenses. For those convicted of less serious offenses, counties were required to pay anywhere from 50–100 percent of the custody costs to the state. Subsequent legislation passed in 1998 capped the maximum annual per-ward payment from the counties to $31,200. Nonetheless, for all juvenile commitments, and especially for the commitment of youths convicted of less serious offenses, the increased cost to counties created by the reform was substantial. Shifting the costs of incarceration to counties, where incarceration decisions were made, caused an immediate and sustained drop in admissions to the California Youth Authority beginning in 1997. Further analysis showed little corresponding increase in incarceration in adult facilities, a greater propensity to divert cases to an alternative non-incarceration resolution, and no measurable effects on reported Part 1 crimes nor on youth arrests (Ouss, 2020).
More recent reforms in adult corrections illustrate the effects of aligning the cost incentives faced by counties with the actual cost of admitting people to prison. Under pressure from a federal court to relieve prison overcrowding and a U.S. Supreme Court ruling that overcrowding in California’s prisons violated the Eighth Amendment rights of state prisoners, California enacted broad corrections reform legislation under the banner of corrections realignment in April 2011, with implementation beginning on October 1, 2011. The legislation halted the practice of revoking people under parole supervision back to prison for technical violations and diverted people convicted of many nonserious, nonviolent, nonsexual offenses to jail sentences or an alternative sanction to be served through some form of community corrections.
This realignment led to a relatively quick reduction in the California prison population, caused by a sharp and immediate reduction in prison
admissions coinciding with the implementation of the legislation. By the end of 2011 (three months into the implementation of reforms) the prison population had declined by roughly 13,000 (an 8% decline). By May 2013, the prison population had declined by nearly 28,000 relative to September 2011 (a 17% decline). In terms of incarceration rates, by the end of 2012 California’s incarceration rate stood at 348 per 100,000, a rate comparable to the rate in 1992, prior to the passage of the state’s tough “Three Strikes” sentencing reform. This is in comparison to an incarceration rate on the eve of realignment’s implementation of 426 per 100,000. The large decline in incarceration had minimal impact on California crime rates (Lofstrom and Raphael, 2016). In fact, despite subsequent reforms that reduced the prison population even further in subsequent years, crime rates continued to decline, with murder rates dropping below the national average for the first time in years in 2014. Little is known about longer-term effects, say a decade after the original policy change, and the committee sees this is as an important topic for further research.
Taken together, the implication of reforming fines and fees and California’s realignment is not so much that financing and criminal justice decisions need to be aligned at the same level of government. After all, budgets and decision making had been aligned at the same level of government in Ferguson and in many county-court systems that levy fines and fees. Instead, the principle of parsimony is served when criminal justice decisions are taken with attention to the true cost of those decisions, both fiscally for the jurisdiction and socially for the communities that such decisions are intended to serve.
Costs and benefits have often been treated asymmetrically in criminal justice policy making and individual decision making, with the costs to government budgets and to communities often being overlooked. The implications of weighing costs are wide-ranging. At the level of statutes and penal codes, changes in sentencing could be scored in a manner similar to spending bills in state and federal legislatures. Policy researchers have also suggested that changes in sentencing policy include racial impact statements that account for the social costs of policy change.
Shifting Policy Approaches to Drugs and Violent Crime
The increase in incarceration rates in the four decades since the early 1970s largely resulted from changes in criminal justice policy (NRC, 2014; Raphael and Stoll, 2013). The proximate causes of the growth in imprisonment rates were an increase in prison admissions per arrest and an increase in the time served in prison (NRC, 2014). In state prison systems, tougher drug sentences accounted for one-fifth of incarceration growth from 1984 to 2004, while tougher sentences for violent crime account for roughly half
(Raphael and Stoll, 2013, p. 27). In federal prison systems, the growth of incarceration was driven primarily by tougher sentences for drug offenders and other public-order crimes. Across both the federal and state systems, legislatively driven policy changes largely drove incarceration grown in the latter decades of the 20th century (Raphael and Stoll, 2013).
Such policy changes largely took the form of state and federal sentencing policy that expanded the scope of criminalization and enhanced sentences for many offenses. Particularly relevant to the current discussion of policy change, the wars on drugs and crime involved the adoption of new rules and routines across many domains, from policing to corrections. These changes in criminal justice policy had large effects on racial inequality in criminal justice involvement. For example, drug arrest rates climbed sharply through the 1980s and are estimated to have contributed to about a third of the rise in imprisonment rates in the 1980s and 1990s (and likely to larger proportions of the increase in the population of people with prior felony convictions and prison time served; Blumstein and Beck, 2012). Increased policing and prosecution of drug offenses was also associated with increased racial disparities in both arrests and imprisonment.
Moreover, most states moved to determinate sentencing regimes, adopted mandatory minimum sentences, or enacted truth-in-sentencing laws that required individuals who are incarcerated to serve a specified minimum percentage of their original sentence. Some states passed repeat-offender statutes, such as “three strikes” laws (Raphael and Stoll, 2013). Nearly half of U.S. states passed sentence enhancements for second- or third-time violent felony convictions, with California’s three-strikes law being particularly punitive (Chen, 2008). Following the lead of many states that passed “truth-in-sentencing” standards, in 1994 Congress passed the Violent Crime Control and Law Enforcement Act, which supplied federal funds for prison construction to states that required people to serve 85 percent of the sentence prescribed by statute (Ditton and Wilson, 1999). These examples indicate a broad shift toward more punitive treatment for offenders across every major crime category (Neal and Rick, 2016, p. 38). Since arrest rates for Black people are higher than for White people, this shift to more punitive sentencing policies was particularly devastating for Black communities (Neal and Rick, 2016, p. 38). Racial disparities in imprisonment are relatively large for drug and violent crimes, and these offenses account for a large share of the prison population. Tonry (2019, pp. 11–13) argues that changes in sentencing policy in these two areas have been important for the overall racial disparity in incarceration. In addition, in the federal system, the shift toward more punitive sentencing for drug crimes affected Black individuals more than White individuals (and drug of-fences accounted for a significant portion of growth in the federal prison population; Neal and Rick, 2016, p. 38).
According to Raphael and Stoll (2013, p. 28), “Collectively, these policy changes have driven the enormous increase in state incarceration rates since the mid-1970s. The increase in the federal prison incarceration rates can be traced largely to the incarceration of drug offenders and the minimum sentences mandated in federal sentencing guidelines.”
The rate of incarceration for drug-related crimes has declined since around 2008, partly the result of drug policy reform over the last two decades that has revised sentencing statutes, policing practices, and prosecution. Sentencing reforms for drug offenses, enacted since the early 2000s, have focused on mandatory sentences (Mauer, 2011b), greater leniency for possession, and widespread decriminalization and legalization of cannabis (Beckett and Brydolf-Horwitz, 2020; Mikos, 2020). A number of jurisdictions have passed legalization statutes that provide for the legal cultivation and sale of cannabis (Mikos, 2020; Kreit, 2016), and in 2013 a Department of Justice memorandum advised federal prosecutors not to interfere with state marijuana legalization laws.
Decriminalization and legalization has been associated with large declines in arrests for cannabis possession (Plunk et al., 2019). Since 2000, the national-level decline in incarceration for drug offenses has been accompanied by a large absolute and relative decline in racial disparity. The prison population incarcerated for drug offenses has declined greatly for Black men and women, although it increased slightly for White people (Table 8-1). Of the total decline in the Black prison population (from 562,000 in 2000 to 399,000 in 2019), about 60 percent is attributable to the decline in the population convicted of drug offenses. Examining federal courts, Light (2022) also finds that racial inequality in length of sentencing between Black and White defendants decreased to less than six months from 2009 to 2018; among those convicted of drug offenses the Black/White gap decreased to zero over the same period. Light (2022) finds these trends to be driven by decreasing sentences for Black individuals and increasing sentences for White individuals, changes in observable case characteristics, and shifts in the prosecutorial use of mandatory minimums.
Like drug offenses, large racial disparities accompany arrest and incarceration for violent offenses. Unlike drug policy, however, sentencing for violent offenses has not become markedly more lenient in recent years, and racial disparities in incarceration for violence have not declined as much as for drug offenses. However, there is some movement in this direction as half of states have now banned life without the possibility of parole for juvenile offenders (a sentence generally meted out for serious violent offenses).7 Moreover, the state with the harshest three-strikes law, California, narrowed the range of offenses for which a third felony carries a 25-to-life
7 See https://www.sentencingproject.org/publications/juvenile-life-without-parole/
TABLE 8-1 State Prison Population by Race/Ethnicity and Offense, 2000, 2019
NOTE: The substantial increase in Hispanic state prison population for violent and property crime offenses observed in Table 8-1 is largely related to the growth of the Hispanic population in the nation as a whole and possible due to changing measurement of this population from 2000 to 2019 (see Beck and Blumstein, 2018).
SOURCE: Data from Sabol and colleagues (2009), Table 7; Carson (2021), Table 15.
sentence. The incarcerated population convicted of violent offenses also declined among Black people, but the reduction in imprisonment of nearly 20,000 contributed only about 12 percent of the decline in Black imprisonment from 2000 to 2018 (see Table 8-1).
Policy approaches to drugs and violence represent two distinct pathways for the criminal justice system. The liberalization of drug policy in the last two decades shrank both enforcement efforts and incarceration rates for the possession and sale of narcotics. Reducing drug arrest and prison admission rates, in turn, has been associated with a large decline in the incarceration of Black men and women. Nevertheless, the enduring pattern of long sentences for violent crimes has sustained racial disparities in incarceration. The two trends in policy suggest that alternative responses to violence that rely far less on long sentences would likely yield large reductions in racial disparity in incarceration.
Research reviewed in Chapters 2 and 4 indicates that there remain significant racial disparities in police stops and searches, arrests, and use
of force. In this section we consider research evidence on policing reforms that could reduce disparity. Racial disparities in policing are often seen as the outcome of two distinct processes (Pryor et al., 2020). First, there are policies and crime patterns that shape the context of police decision making and deployment. For example, there may be a more visible police presence in Black neighborhoods due to explicit enforcement strategies such as hot spot policing, responding to more emergency calls for service in Black neighborhoods, or differential policing employed in predominantly minority cities. This differential presence will tend to produce more contact between Black residents and police, regardless of the individual dispositions of officers. Second, officers may vary in their dispositions, and some may use their discretion to treat some racial groups differentially. This differential treatment may result in differential scrutiny of Black pedestrians and drivers, differential and harsher treatment during common police-citizen interactions, and a greater volume of stops and interactions that result in no actions taken by police officers.
Changing the Context of Policing through Oversight and Accountability
In this section we describe large-scale changes in the deployment of police resources obtained through civilian and Department of Justice oversight efforts. In some cases below, as in New York City’s abandonment of an unconstitutional stop-and-frisk policy, racial inequality was the focal point of court oversight. In other cases, such as Washington State’s experience with pretextual traffic stops, court oversight narrowed police discretion, and racial disparity declined as a consequence.
Notable recent cases of court oversight have focused on so-called Terry stops, named for the landmark U.S. Supreme Court decision in Terry v. Ohio. In that case, the court held that a police officer can temporarily detain a citizen if the officer has a reasonable, articulable suspicion that the detained person was engaged in criminal activity. The court also ruled that a citizen can be superficially searched (frisked) if the officer has a reasonable, articulable suspicion that the citizen is armed and dangerous. Guided by the broken windows theory that policing of low-level offenses serves to limit the perpetration of more serious crime, stops and searches became an important policing tool in many jurisdictions in the 1990s.
Policy developments and court action in New York City (NY) have been a key focus for researchers studying the connections between policing and racial inequality. In New York City, the New York Police Department (NYPD) had, since the early 1990s, embraced order-maintenance policing
and enforcement actions against low-level offenses. The policy is reflected in the increased number of pedestrian stops and low-level arrests from the 1990s through the first decade of the 2000s. In 2003, the number of stops totaled around 150,000, and by 2011 it had climbed to nearly 700,000. Litigation around stop-and-frisk in New York began in 1999 in a class action lawsuit, which alleged that the NYPD targeted individuals on the basis of race and national origin without reasonable suspicion, in violation of the Fourth and Fourteenth Amendments (Daniels v. City of New York).8
A settlement in the case required, among other provisions, that the NYPD maintain an anti-racial profiling policy, engage in anti-profiling training, and maintain a database of stops. The settlement did not include an independent monitor, and police compliance with the settlement appears to have been poor. A second case was brought, Floyd et al. v. City of New York et al., in which the court ruled in 2013 that the NYPD was continuing to engage in unconstitutional stop-and-frisk practices that were racially discriminatory, targeting Black and Latino New Yorkers. Among the remedies, an independent monitor was appointed to oversee policy changes that would bring the NYPD back into compliance with the law. The court also required body-worn cameras to be used in a number of police precincts and that community consultation be adopted for the process of policy change.
Immediately following the Floyd ruling, the number of stops, frisks, and arrests dropped precipitously. The top 10 precincts recorded 685,724 stops in 2011, compared to just 45,788 stops in 2014 (Table 8-2). The 93 percent reduction in stops in these precincts caused large declines in racial disparities in stop rates. While the differences among Black, Hispanic, and White New Yorkers in their likelihood of being stopped, questioned, and frisked by the police declined precipitously, the racial composition of the much smaller pool of post-Floyd stops is similar (Black New Yorkers accounted for 53% of all stops in 2014 in comparison to 51% of all stops in 2011). Among Black men and women, police stops in the top 10 precincts declined by 326,092 in three years, far exceeding the decline for any other racial group. We also observe a large decline in stops for Hispanic New Yorkers (White et al., 2016). This is an example of a reform that has disproportionately positive impacts on affected populations even when relative disparities are not reduced by much.
In these cases, a large across-the-board reduction in stops, frisks, and arrests, in a context with large preexisting racial disparities, greatly reduced police contact among Black and Hispanic residents. Court oversight resulting from private litigation in New York coupled with independent
8Daniels v. the City of New York, 99 Civ. 1695 (SAS) (S.D.N.Y. 2001). See https://casetext.com/case/daniels-v-the-city-of-new-york-sdny-2001
TABLE 8-2 Numbers of Stops, Frisks, and Arrests in the Top 10 Precincts in 2011, When Floyd v. City of New York Decided That the New York Police Department Policy of Stop and Frisk Was Unconstitutional, and in 2014, by Race
SOURCE: Data from White and colleagues (2016).
monitoring of unconstitutional policing significantly reduced absolute but not relative disparity in Terry stops, police searches, and arrests.
Pretextual Traffic Stops
Related litigation has focused on pretextual traffic stops. A pretextual stop occurs when an officer identifies an objective violation of traffic law and lawfully stops the motorist, but the officer’s actual intention is to investigate a hunch that, by itself, would not amount to a reasonable suspicion or probable cause. The 1996 U.S. Supreme Court Case of Whren v. United States9 ruled that the violation of a traffic law renders the stop reasonable
9 517 U.S. 806 (1996). See https://supreme.justia.com/cases/federal/us/517/806/
under the Fourth Amendment, even though the true purpose of the stop, absent the traffic violation, would not be constitutional. Legal scholars argued that the Whren decision greatly enlarged police discretion in a way that allowed them to target motorists of color (Rushin and Edwards, 2021; Chin and Vernon, 2014; Maclin, 1998; Harris, 1997). Pretextual stops, like Terry stops, are a first point of contact generating vast numbers of interactions between police and the public.
Although state constitutions and state law may establish stronger protections against pretextual stops, few states have strayed from the core holding in Whren (Lawton, 2015). Unusually, the Washington State Supreme Court in 1999 imposed a ban on pretextual stops as violating the state constitution’s provision that “no person shall be disturbed in his private affairs, or his home invaded, without authority of law” (Rushin and Edwards, 2021, p. 654). Thirteen years later, in 2012, another state supreme court effectively repealed the ban in Washington State by narrowing the restrictions on pretextual stops. Although racial profiling was not central to the Washington State Supreme Court decisions, in contrast to the Floyd case in New York, and neither decision was directly intended to affect racial disparity, a large body of legal scholarship has examined racial profiling by police in traffic stops, particularly since the Whren decision. Did the 13-year ban on pretextual stops in Washington State reduce racial disparity in traffic stops?
Rushin and Edwards (2021) studied the impact of legal changes in Washington State, comparing the number of stops of non-White and White motorists in the period during the ban to the number of stops made in the later period of its relaxation. In a difference-in-difference analysis of race × county × month counts of stops, non-White drivers experienced an additional 250 stops each month in a county, following the relaxation on the ban on pretextual stops. The average number of stops among Black and Hispanic drivers was approximately 2,500 each month. Racial disparity declined in night-time stops under the “veil of darkness” (Ridgeway and MacDonald, 2009). The analysis provides rare evidence of the causal effect of pretextual stops on racial disparity and the effectiveness of a legal ban on reducing disparity. Relying only on observational data, however, other changes in the legal and political environment that coincide with the relaxation of the ban on stops could threaten causal inference. Around the time of the ban, marijuana was legalized in Washington, a new governor and attorney general were elected, and Republicans won the state senate. Causal inference would be biased if one of these changes induced racially disparate changes in police behavior. Although it is unclear if changes in drug policy or state politics affected racially disparate police behavior, it is clear that that marijuana legalization coincided with a significant reduction in stops for Washington motorists in general. Similar to the reduction in the national imprisonment rate for drug offenses, the reduction in traffic
stops accompanying marijuana legalization likely disproportionately reduced stops among non-White motorists, even as stops increased among minority motorists when the court narrowed the ban.
Federal Pattern or Practice Investigations
In a third example of changing police oversight, the Department of Justice since 1994 has expanded scrutiny of police departments through “pattern or practice” investigations. Under Section 14141 of the Violent Crime Control and Law Enforcement Act,10 the Attorney General is authorized to initiate structural reform litigation against police departments suspected of systemic misconduct that “deprives persons of the rights, privileges, or immunities secured and protected by the Constitution or laws of the United States” (Stuntz, 2006, p. 798). This authorization, called the pattern or practice provision, is considered “the most important legal initiative of the past twenty years in the sphere of police regulation” (Stuntz, 2006, p. 798).
Investigative work under Section 14141 has been undertaken by the Special Litigation Section of the Justice Department’s Civil Rights Division, which has specified a number of goals for the program, including ending unconstitutional policing by curbing officer use of force and curbing discrimination based on race, ethnicity, gender, or sexual orientation in an effort to establish “bias-free policing” (Walker, 2022). From 1994 to 2017, the Department of Justice initiated 69 investigations of local police departments, reaching settlements with 40 departments, involving 20 judicially enforced federal consent decrees (Walker, 2022). Settlements typically involve new procedures for investigations into use of force, searches and arrests, citizen complaints, and officer disciplinary actions. Several papers report comparative case studies of consent decrees in several sites. Rushin (2015, p. 1422) concludes that the “available evidence suggests that [systemic judicially enforced police reform] has been an effective tool for reducing misconduct in several police agencies.” Chanin (2015) reviews pattern or practice investigations in Pittsburgh (PA), Washington (DC), and Cincinnati (OH) and generally finds evidence of reduced police misconduct, measured by statistics on use of force, citizen complaints, and disciplinary action in response to alleged misconduct.
In Los Angeles (CA), Stone and colleagues (2009) documented increased public satisfaction and decreased frequency of serious levels of force among the Los Angeles Police Department (LAPD) under a consent decree following the Rampart scandal. A consent decree in Newark (NJ)
10 42 U.S.C. § 14141 (1994; re-codified as 34 U.S.C. § 12601). See https://www.justice.gov/crt/conduct-law-enforcement-agencies
was associated with fewer police stops among Black and Latino residents, although spatial concentration of field inquiries remained (Chillar, 2022). Walker’s (2022) review of consent decree evaluations indicates reductions in use of force and sustained improvements in department operations, though many sites have gone unevaluated and there has been no general evaluation of the Department of Justice pattern or practice program.
Resistance to Change in Police Culture
Some researchers have argued that consent decrees have changed police behavior, but it has been more difficult to bring about lasting change to police organization and culture. A review by Alpert and colleagues (2017) reports that although consent decrees are often found to be effective while federal monitors are in place, temporary oversight has not produced enduring change. For example, qualitative data from Pittsburgh show that police officers met the 1997–2002 consent decree with resistance and negativity (Davis et al., 2005). The city implemented new systems to track the use of force, traffic stops, and searches, and established new procedures for officer accountability, new policies, and training. Officers noted increased accountability and oversight mechanisms during the period of the consent decree but also reported that changes in management made them less aggressive in fighting crime. The reforms remained intact one year after most of the consent decree requirements were lifted (Davis et al., 2005). Without additional provisions for enduring oversight, “there is no established mechanism or requirement to maintain any of the accountability features that were established under the consent decree” (Alpert et al., 2017, p. 243).
While there is evidence that federal oversight has reduced police misconduct, particularly while a court monitor was in place, oversight has often been resisted by police leadership and unions (Walker, 2022; Rushin, 2017). Police objections to oversight are well documented, and police leadership and middle management have regularly opposed and refused cooperation with federal monitors (Walker, 2022). Increasing the obstacles to effective oversight, police unions and their collective bargaining agreements frequently shield officers from public scrutiny. A review of police union contracts in 178 municipalities identified seven provisions that shield officers from accountability. These include the delay of investigative interviews in cases of police misconduct, limitations on the use of officers’ disciplinary history in misconduct hearings, time limits for imposing discipline, and independent arbitration that regularly overturns terminations and reduces suspensions (Rushin, 2021, 2017). There is also evidence of countermovement organization among police and supporters in competitive opposition to protests for police reform (Solomon and Martin, 2019).
One line of research has examined whether police resistance to oversight and accountability has resulted in a withdrawal of police services that ultimately causes an increase in crime. Devi and Fryer (2020) examined this possibility in an empirical analysis of the effects of federal and state pattern or practice investigations on policing behavior and crime rates in 27 pattern or practice investigations. In 22 out of 27 cases, investigations are estimated to reduce overall crime rates, and to reduce homicides on average by 61 per investigation over 24 months following the investigation. In five of 27 cities, however, investigations were initiated following high-profile killings of Black citizens by police that sparked protests and gained national media attention. These “viral” incidents in Baltimore (MD), Chicago (IL), Cincinnati (OH), Riverside (CA), and Ferguson (MO) were followed by large increases in homicide and overall crime rates. An estimated 1,099 excess homicides and 31,293 excess felony crimes were recorded over the two-year period following an investigation in cities with a preceding viral incident. That being said, it is not clear whether the pattern or practice investigation or the high-profile incident preceding the investigation caused the additional crime. Evidence of declining police stops in four of the viral cities suggests that reduced police activity following the announcement of the federal investigation drove the increase in crime in those localities. Increasing crime as a result of reduced policing in response to oversight efforts was also reported in Los Angeles (Prendergast, 2021) and Cincinnati (Shi, 2009). In these cities, efforts to increase police accountability did not include close monitoring of police activities aimed at preventing and solving crime. The reduction in police services that accompanied federal oversight was associated with an increase in crime, including a jump in homicide rates in Los Angeles. In Los Angeles, Prendergast (2021) observed a 40 percent decrease in the arrest-to-crime rate and an increase in homicides (2000–2003) for the LAPD, and no corresponding changes in an adjacent department. The impacts were pronounced in predominantly Hispanic neighborhoods.
Related research has studied the so-called Ferguson Effect, where policing activities may have been reduced following the social protests and Justice Department investigation in Ferguson following the police killing of Michael Brown. Premkumar (2019) finds a decline in low-level arrests following protests in Ferguson that he associates with increased public scrutiny of the police, rather than the federal investigation, and an increase in homicides. Against this finding, Rosenfeld and Wallman (2019) find little evidence of de-policing measured by arrest rates, given homicides across 53 large cities following the killing of Michael Brown (Rosenfeld and Wallman, 2019).
Community-led efforts to reduce racial disparities in policing have focused on increasing social accountability measures through
non-department oversight and engagement. An analysis of civilian oversight agencies in the period 1980–2014 found lower levels of racial disparity in the disorderly conduct arrests where oversight boards had broad authority to conduct investigations, although racial disparities in policing outcomes tend to grow as the oversight agency ages (Ali and Pirog, 2019). The Department of Justice’s Office of Community Oriented Policing Services led a report on reconciliation efforts to acknowledge past harm in law enforcement practices while using historical context to inform solutions to reduce racial profiling and disparities (Mentel, 2012). Public scrutiny has led to a national reconsideration of the role of police officers and the responsibilities that should fall under their discretion. Judicial consent decrees are an example of efforts designed to reduce racial disparities with only evidence of temporary success.
One area of police operations that aims, in part, to improve police accountability involves the use of body-worn cameras. A significant literature now examines the effects of body worn-cameras on police behaviors. Body-worn cameras are intended to document interactions between police and citizens to improve transparency and accountability. The main mechanism by which cameras might affect behavior concerns the deterrence of wrongdoing. The effects of these cameras may be experienced by both police and the citizens with whom they interact. Hopes that body-worn cameras might reduce police misconduct were reflected in the recommendations of President Obama’s Task Force on 21st Century Policing (2015, pp. 31–32) and the Leadership Conference on Civil and Human Rights (2015, Principle 4).
Lum and colleagues (2020) provide a thorough review of studies on this subject conducted between 2004 and 2018, examining the effects of body-worn cameras on outcomes including officer use of force, citizen complaints, arrests, and pedestrian and traffic stops. Their review examines 30 studies, 17 of which are known to have been fielded in the United States. Across a range of outcomes, the review indicates there is no consistent evidence that body-worn cameras are associated with either less or more enforcement action, officer injuries, or officer use of force. Against this general pattern of findings, however, body-worn cameras were found to be associated with significant reductions in complaints against officers and a significant increase in non-traffic citations. Given the relatively small body of research, meta-analysis results are sensitive to the included studies. M.C. Williams and colleagues (2021) add two more recent studies to the 30 examined by Lum and colleagues (2020) and report stronger evidence for a reduction in officer use of force.
The effect on racial disparities has been examined in New York City, where body-worn cameras were introduced following the Floyd case on stop-and-frisk. A randomized controlled trial evaluated the effects of cameras in an experiment involving 40 police precincts and 3,889 NYPD officers (Braga et al., 2021). In the design of the experiment, the 40 precincts were sorted into 20 pairs matched according to the number of complaints made to the Civilian Complaint Review Board, a civilian oversight body. Body-worn cameras were then assigned randomly to one precinct within each pair, and a random sample of stops was drawn for the one-year intervention period. The study found that body-worn cameras were not significantly associated with the numbers of arrests, arrests with force, summons, or crime complaints. However, officers in the treatment group experienced 21 percent fewer review board complaints and a 38 percent increase in the number of stops. An audit of a random subset of stops showed that New Yorkers in the treatment category were 10 percentage points more likely to be Black. Stops made by the body-worn-camera group of officers were significantly more likely to be evaluated as not meeting constitutional justifications for stops, frisks, and searches by the study team. The authors conclude both that the NYPD still has problems with unconstitutional stops of citizens made by its officers and that the presence of the body-worn cameras may be enhancing officer compliance with NYPD policy directives to document citizen stops (Braga et al., 2021).
Major efforts to change the context of policing to reduce racial inequality have been mounted through measures to expand judicial, federal, and public oversight. Significant reductions in the exposure of Black and Latino citizens to police contact have been obtained by regulating police discretion at the point of first contact between police and the public—in pedestrian stops and traffic stops. Evidence from New York City and Washington State shows that the elimination of stop-and-frisk and pretextual stops greatly reduced interaction between police and people of color. In New York City, relative racial disparity did not decline but the great reduction in police activity had substantial effects in minority communities. Pattern or practice investigations leading to Department of Justice settlements have been associated with reduced use of force, fewer citizen complaints, and declines in crime, but these effects are short-lived. Research on body-worn cameras shows that in some jurisdictions technological changes to improve accountability can reduce the use of force and citizen complaints against the police. However, other jurisdictions have obtained no benefit from body-worn cameras, and on average the results are not significantly positive.
Reducing racial inequality by improving oversight and accountability is often met with staunch resistance from police leadership, middle management, and line officers, although the evidence indicates this has more clearly stymied the implementation of reform rather than resulting in a withdrawal of police services. In some areas, particularly related to federal oversight of local police departments, research suggests that methods for accountability and oversight must be strengthened because constitutional compliance is short-lived, and in some localities police have withdrawn services, resulting in increases in crime.
Changing the Disposition of Officers
Whereas changes in oversight have affected the context of police decision making, direct efforts to change the use of discretion have focused on training and hiring policy. Recent training efforts directly aimed at racially unequal policing have focused on implicit bias. The idea of implicit bias rests on a social-psychological theory in which an individual’s judgments and decisions are based on heuristics that simplify cognitive tasks to enable a speedy resolution. Stereotypes can inform judgments, even if the decision maker would reject the stereotype in a more deliberate decision process. This rapid decision making in combination with stereotypes is the basis for implicit bias.
As the historical discussion in Part I showed, longstanding and deep-seated stereotypes associate Black people with criminality and violence. Similar findings emerge from contemporary psychological research, in which experimental subjects are found to associate Black people with fears about dangerousness, threat, violence, and irredeemability (Donders et al., 2008; Eberhardt et al., 2004). Research shows that these deep-seated associations are just as prevalent among police officers as they are among the public. Stereotypes that associate people of color, and especially Black people, with violence and other criminality may lead police officers to act—perhaps even unconsciously—with bias.
Many police departments have responded to national pressure to address disparate outcomes by reforming officer training. Officer training has evolved to better equip officers in the field, including in their decision making processes. Implicit bias trainings aimed at reducing racial profiling (and therefore disparities) have been widely implemented in police departments throughout the country. There are few rigorous evaluations of the effects of such anti-bias training on racial disparities. Evaluations have found that trainees showed greater awareness of personal bias and a better understanding of how bias contributes to racial disparities. Still, there is no rigorous evidence base indicating that anti-bias training reduces racial disparity. For example, Worden and colleagues (2020) evaluated an implicit
bias program for NYPD officers. They found that while the training had a moderate effect on officers’ knowledge about implicit bias immediately after the training, a follow-up survey revealed a decay effect in terms of knowledge gained. Moreover, the authors report that they found insufficient evidence to conclude that racial and ethnic disparities in police enforcement actions were reduced because of the training.
Recently, some jurisdictions have passed legislation focusing on eliminating explicit bias from police departments, largely through enhanced screening during the hiring stage and by making evidence of explicit bias a separate articulated element of background investigations. For example, California’s recently passed AB846 requires background investigators as well as psychological examiners to include explicit bias as a review criterion for new recruits.11 While the law has yet to be implemented, the legislation will likely incorporate a routine review of past statements and social media activity for evidence of explicit bias.
The value of increasing diversity in the police ranks is supported by research showing that female police and people of color engage in less enforcement activity than male and White officers (Ba et al., 2021). Unusually for research on the effects of police diversity, Ba and colleagues analyze highly disaggregated data that allow comparisons of officers, who vary by race and gender and who are patrolling the same beats at the same time of year and at the same time of day. Earlier research had studied coarse geographic units (Legewie and Fagan, 2016; Donohue and Levitt, 2001). Analyzing data from the Chicago Police Department, they find that Black officers make fewer stops and arrests compared to White officers in the same settings. Notably, Black officers make significantly fewer stops for “suspicious behavior.” The reductions in enforcement activity are largely due to fewer enforcement actions taken against Black citizens. The authors caution that further diversification of the police department is likely to have complex and heterogeneous effects, although they conclude that “the results strongly show that diversification can reshape police-civilian encounters” (Ba et al., 2021, p. 701). Similar results are provided in an analysis by Hoekstra and Sloan (2022) of 1.6 million 911 calls. They find that in Black neighborhoods (at least 80% Black), at the level of census blocks, White officers are about five times more likely to use gun force than Black officers. Another study exploits 25-year panel data on the race of U.S. sheriffs and shows that the ratio of Black-to-White arrests is significantly higher when a White sheriff is managing a department (Bulman, 2019). Relatedly, Harvey and Mattia (2022) found an association between the successful litigation over racially discriminatory practices within law enforcement agencies and
11 See https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=201920200AB846
a reduction in absolute and relative crime victimization among Black people without a corresponding increase in victimization among White people.
Implicit bias training, hiring procedures that foster diversity, and background checks for explicit bias can also be viewed as elements of a larger organizational context that promotes a department-wide culture of fairness and accountability. From this point of view, eliminating bias and building legitimacy is as much a challenge of culture change and organizational design as a challenge of shifting individual psychology (Tyler, 2017; Fridell and Lim, 2016). There is more research on individual police decision making (see, for example, Owens et al., 2018) than studies of multifaceted efforts at cultural and organizational change aimed specifically at reducing racial inequality, and the committee views this as an important area for future study.
Racial Inequality and Police Deterrence Tactics
Police tactics to deter crime have important implications for racial inequality in the criminal justice system. Deterrence-based tactics have been central to the evolution of crime-prevention strategies and a focal point for researchers. Chalfin and McCrary (2017) observe that deterrence-based strategies often involve the police in signaling the threat of sanctions, so policing necessitates close contact with the public. Deterrence-based tactics can also concentrate policing in predominately Black and Latino communities, where the risks of crime are assessed by police to be highest. Finally, deterrence-based tactics can also widen police discretion to stop and search the public. Due to close contact with the public, the concentration of resources in disadvantaged communities, and widening discretion, deterrence tactics may also have disparate effects in Black, Latino, and Native American communities.
The large literature examining the effects of police force size and policing tactics on crime consistently finds that higher staffing levels reduce crime rates with statistically and substantively significant impacts on homicide (the National Academies of Sciences, Engineering, and Medicine, 2018, see Chapter 7; Chalfin and McCrary, 2017). While fewer studies examine the impact on racial inequality, one recent paper analyzes the relationship between race-specific crime rates and the number of police officers in 242 large U.S. cities from 1981 to 2018 (Chalfin et al., 2020). The results indicate that on average, an increase in police staffing is associated with a reduction in homicide victimization, with the effect on homicides per 100,000 being twice as large for Black homicide rates relative to White homicide rates. The authors also find that higher police staffing levels lead to a significant decrease in index crime arrests (and presumably index crimes), with the decline in index crime arrests four to six times larger on a per-capita basis for Black suspects relative to White suspects. The authors interpret
this finding in particular as strong evidence of a deterrence effect of policing and evidence of “double-dividend” in the form of both less crime and fewer arrests for the most serious offenses. However, the authors also find that the effects on homicide are smaller in cities with a higher percentage of Black residents. Moreover, analyzing quality-of-life arrests, the burden of low-level arrests found to be associated with an additional police officer is 70 percent greater in the Black population than for White people. The authors interpret the race-specific results as indicating that additional policing presents a mixed bag of benefits and costs, with African Americans experiencing the largest per-capita declines in homicide and Part 1 offenses, as well as declines in Part 1 arrests, but increases in quality-of-life arrests. Moreover, their estimates of the interaction terms between policing and the proportion of Black residents suggests that there are some predominantly Black cities in the country where the marginal benefits from having additional police are small if not zero.
In contrast to the broad brush of low-level enforcement and a blanket police presence, a highly concentrated deterrence strategy, called focused deterrence, narrowly targets police attention on a relatively small number of people who are identified as at high risk of crime and victimization (the National Academies, 2018). Focused deterrence shares some attributes of community-based anti-violence strategies reviewed in Chapter 6. In contrast to these approaches, however, focused deterrence relies on the threat of serious formal sanctions from the criminal justice system as a tool to motivate behavior change among individuals and groups. Focused deterrence is a place-based strategy that aims to intervene with specific individuals or gang networks in an effort to head off serious crime before it occurs. As a police-based deterrence tactic, focused deterrence involves the direct communication of the consequences of continued criminal offending and the availability of social services to targeted subjects. Often this communication takes the form of a “call-in,” a “forum,” or a “notification,” involving face-to-face meetings with police, service providers, and community representatives. Leading examples—such as Operation Ceasefire, which was first fielded in Boston and later replicated in Newark, Los Angeles, and other cities—are typically deployed in highly disadvantaged minority communities (see Box 8-1 for a description of Operation Ceasefire; see also Chapter 6).
The most successful focused deterrence programs are associated with significant reductions in shootings and other serious violence (Braga et al., 2018). In 1996, Operation Ceasefire in Boston reported a 63 percent reduction in youth homicide, and a 44 percent reduction in youth gun assaults in one high-risk district (Braga et al., 2018). However, these declines occurred
as violent crime was also decreasing in other cities, including cities that did not adopt an Operation Ceasefire approach. Operation Peacekeeper in Stockton (CA) was associated with a 42 percent reduction in gun homicide (Braga et al., 2018). Ceasefire in Rochester (NY) was associated with a 25 percent reduction in homicides with Black male victims ages 15–30 and a 27 percent reduction in gun robberies for the same group; however, the total homicide rate and the gun assault rate were not reduced in Rochester (Braga et al., 2018). In the 24 observational evaluation studies identified by
Braga and his colleagues (2019), focused deterrence was found on average to have a “statistically significant, moderate crime reduction effect.” The National Academies Committee on Proactive Policing reviewed the evaluative literature on focused deterrence and concluded that
[e]valuations of focused deterrence programs show consistent crime-control impacts in reducing gang violence, street crime driven by disorderly drug markets, and repeat individual offending. The available evaluation literature suggests both short-term and long-term area-wide impacts of focused deterrence programs on crime. (the National Academies, 2018, p. 175)
A literature scan of focused deterrence programs from the Urban Institute Justice Policy Center uncovered the following themes in their implementation: the need for analysis-driven goal setting, the importance of strong partnerships with local communities to build trust and a sense of joint ownership, challenges with replication that require tailoring programs to the local context, and the need for consistent monitoring and reevaluation for sustainability (Matei et al., 2022). Qualitative researchers report that the success of focused deterrence depends on a belief in the legitimacy of the police among local community residents and a procedurally just approach to the call-in meetings by police (Braga et al., 2018).
A related set of strategies in which police resources are deployed in relation to an assessment of risk involves directed patrol at crime hot spots. Directed patrols aimed at gun violence deploy police in small areas where shootings have been concentrated. In some research sites, gun patrols were preceded by police outreach to local households informing residents of upcoming patrols. The patrols themselves typically involve proactive car and pedestrian stops aimed at gun interdiction. Like focused deterrence, directed patrol has been found to be effective at reducing gun violence, and experimental results indicate reductions in gun-related assaults, robberies, and homicides (the National Academies, 2018; Koper and Mayo-Wilson, 2006). As the National Academies report on proactive policing discusses, strategies like directed patrol that initiate intense police contact in small areas (in practice, often in Black and Latino neighborhoods) have been shown to reduce serious violence but also carry the risk of discriminatory treatment and community mistrust. This finding points to the possibility of a tradeoff between reducing crime and reducing racial disparity.
Such research on the effectiveness of directed patrols cannot be viewed in isolation from the substantial evidence we have reviewed on the discriminatory treatment of minority communities in police stops. As Webster (2022, p. 40) states, “Given the potential for abuse in proactive gun-law enforcement, police must have strong systems of internal and external accountability to ensure that practices are not only legal but minimize
harms and are acceptable to community members.” This observation is particularly relevant at the time of this writing when levels of gun violence and violent victimization have increased sharply among young men during the ongoing COVID-19 pandemic. To meet this challenge, the evidence we have reviewed suggests that time-limited and spatially focused directed patrols, in combination with strong safeguards, may provide one tactic in times and places in which gun violence has escalated. More generally, because police-based deterrence strategies can carry significant risks of racially disparate criminal justice contact, community collaboration (see Chapter 6) and social policy support (Chapter 7) offer important complementary approaches in a comprehensive strategy. Community collaboration has taken the form of civilian oversight, and community consultation and partnership with police. Complementary social policy has taken the form of behavioral health interventions, jobs and training programs, and community-based physical and mental health treatment. Combining police-based deterrence strategies with community partnership and social policy offers an evidence-informed approach to addressing gun violence in a way that protects against the risks of growing racial inequality in criminal justice involvement.
Despite significant evidence, reported in Chapter 4, of racial disparities in court outcomes, little research has been conducted on reforms to reduce racial disparity in court processing. Several of the main efforts to reduce these disparities fall within three categories:
- Pretrial release decisions and risk assessment;
- Mitigation of fines and fees; and
- Modifications of both judicial and prosecutorial discretion, specifically for crimes of poverty.
Many other areas of innovation have been active over the last decades, including specialty courts for specific populations, special procedures for diversion, and alternative approaches to accountability such as restorative justice. Although applications in many of these areas seek alternatives to the punitive approaches that characterized the era of high incarceration rates, programs are often small or lack detailed research that addresses our specific charge of reducing racial inequality.
Pretrial Release Decisions and Risk Assessment
Research indicates significant unwarranted racial disparities in pretrial release decisions (see Chapter 4). Pretrial detention is associated with a
higher probability of a guilty plea and worse sentencing outcomes (Dobbie et al., 2018). Moreover, policy changes that inadvertently reduce racial disparities in pretrial detention have been found to reduce racial disparities in case outcomes (MacDonald and Raphael, 2020). Reducing disparities in pretrial detention thus holds the promise of reducing disparities downstream for sentencing and imprisonment.
Two main and inter-related approaches to reducing unwarranted racial disparities in pretrial processing have involved bail reform and the use of predictive tools to inform judicial decision making. Bail reforms have recently been adopted in New Jersey, New York, Vermont, and Washington (DC), as well as in localities such as Cook County (IL) and Atlanta (GA). Quantitative risk assessment has been widely adopted not just for making pretrial decisions but also in prison facilities and community corrections. Bail reform and risk assessment have often been implemented together, where the range of offenses where bail could be set is reduced and a risk-assessment instrument is introduced to inform the decision about pretrial detention.
A major statewide reform of this kind was adopted in New Jersey in 2017. New Jersey’s criminal justice reform had five main components: (1) a reduction in the use of money bail; (2) the use of a risk tool and a decision process to inform release conditions; (3) a power of pretrial detention without bail; (4) a new pretrial monitoring program; and (5) speedy trial laws that set a schedule for the first court appearance and major milestones such as indictment. New Jersey’s risk-assessment tool was a nine-item inventory based on the defendant’s current age and criminal history that yielded one risk score for the likelihood of committing a new crime and another for the likelihood of failing to appear in court.
An evaluation of this bail reform in New Jersey compared court records in the period prior to the reform with those from the year in which the bail reform was adopted (2017). Bail reform was found to be associated with a reduction in the number of less serious arrests, and with large reductions in the number of jail stays—a 16 percent reduction in the number of jail stays of three days or more and a 36 percent reduction in jail stays of 10 days or more (Anderson et al., 2019). Between 2012 and 2018, the pretrial jail population declined by 44 percent, a decline of 6,000 detainees. This decline included 3,000 fewer Black defendants, 1,500 fewer White defendants, and 1,300 fewer Hispanic defendants, clear evidence of a disparate impact of the reform on Black and Hispanic defendants. Still, in 2018 Black defendants accounted for 54 percent of the pretrial jail population, the same proportion as before the reform (in 2012; Grant, 2019).
In 2018, the Philadelphia (PA) district attorney mplemented an advisory change to end monetary bail for defendants who were charged with certain low-level offenses. Ouss and Stevenson (2020) used a differences-in-differences
approach and found that the policy was associated with a 22 percent increase in the likelihood that a defendant would be released with no monetary or supervisory conditions but had no effect on pretrial detention. The authors found no evidence to support the notion that cash bail or pretrial supervision has a deterrent effect on failure-to-appear in court or pretrial crime.
Risk-assessment instruments for pretrial decision making have been a focal point for policy reforms that aim to reduce racial disparities and jail populations. Although quantitative predictive instruments hold the promise of eliminating judicial bias, as is evidenced by the New Jersey reform, the items comprising a risk tool such as age at first arrest and the number of prior convictions may themselves be subject to racial bias (Harcourt, 2015; Starr, 2015; Holder, 2014). The impact of risk tools on racial inequality thus remains an empirical question. Few jurisdictions have evaluated the validity of their risk tool, studied correlations with race and ethnicity, or modified risk assessment to remove racial correlates from the instrument. The impact of risk assessment on racial disparity thus remains an urgent question for research.
However, at least two jurisdictions, in Minnesota and New York State, have used and modified risk tools with a view to reducing racial disparity and have provided detailed assessments of the effort. The Fourth Judicial District of Minnesota, Hennepin County, has developed a reiterative process of assessing scale validity and mitigating racialized differences in bail decisions. Hennepin County first implemented the use of risk assessment in 1972, adopting and slightly modifying the well-known “Vera Scale.” A 1992 evaluation studied the sensitivity of individual risk factors to the inclusion of race in an equation predicting pretrial failure to appear or risk to public safety. It found that race was significantly predictive, but only two of the seven variables included on the Vera Scale were also significant (Goodman, 1992). A new scale minimizing the importance of community affiliation was implemented in 1992 based on these findings. The 1992 assessment’s validity was measured in 2006; while this tool was found to improve predictive power by three percent compared to the Vera Scale, only five of the nine variables were found to be significantly predictive, and three of the four found not to be predictive were correlated with race. These three factors were removed from the scale as a consequence, and a new iteration of the scale was implemented in 2007 (Podkopacz et al., 2006). The scale was again updated in 2015 and assessed for validity in 2018. Analysis by race showed that White defendants were about 50 percent more likely to fall in the low-risk category than non-White defendants. The failure rate for White people was about 10 percent higher than for non-White people,
although the difference was not statistically significant (Podkopacz et al., 2018).
Like Hennepin County, New York City has also modified its pretrial risk-assessment instrument to reduce racial disparity. In 2003, the New York City Mayor’s Office of Criminal Justice contracted the not-for-profit Criminal Justice Agency to create a pretrial release assessment tool, which it then implemented. In 2019, this tool was evaluated by the agency for disparate assessment by race and updated to reduce disparity. Under the 2003 tool, White defendants were recommended for release at a rate 7.6 percentage points higher than Black defendants (70.3% vs. 62.7%). To reduce racial disparity and create better predictive capacity, the Criminal Justice Agency updated its release assessment. In comparison to the 2003 tool, the 2019 tool significantly increased sum recommendations for release. If this newest version of the tool had been available and implemented with a 2014 sample, it would have recommended an additional 41,600 Black individuals, 19,800 Hispanic individuals, and 8,000 White individuals for release on their recognizance. Moreover, the new tool decreased the disparity in release recommendations for White and Black defendants to 4.3 percent (Luminosity and the University of Chicago, 2020).
There are relatively few academic studies that assess the impact of risk assessment on racial disparity. Kleinberg and colleagues (2018) analyze court data from New York City to assess machine learning predictive tools for pretrial decision making. In contrast to the simple risk-assessment scale, which typically takes a weighted sum of a dozen or so measures of criminal history and current charges, machine learning methods fit complex nonlinear functions of predictive variables on training data, which are then used in the field for out-of-sample prediction. For Kleinberg and colleagues, the input variables were similar to those regularly available to judges at New York City bail hearings and related to current offense and prior criminal history. Machine learning, however, constructs a potentially complex function of inputs to make predictions that balance model complexity and prediction error. The study also exploits the random assignment of judges, so the analysis can compare outcomes for defendants who are otherwise equal but vary in the judges’ release decisions. Kleinberg and colleagues find that a machine learning algorithm will detain Black defendants at a higher rate than White people to reduce the rate of re-arrest or failure to appear. But the algorithm can be tuned to control racial disparity, with only a slight increase in pretrial failure. In short, pretrial failure can be minimized subject to a racial equity goal, but in that case, race must be explicitly taken into account in the pretrial detention decision.
In one of the few empirical studies of the impact of risk assessment on racial disparity, Stevenson and Doleac (2022) analyze data on judges’ sentencing behavior in Virginia (see also Sloan et al., 2018, for pretrial release
in Texas; Stevenson, 2018a, for pretrial release in Kentucky; and Berk, 2017, for parole decisions in Pennsylvania). Virginia courts began using a risk tool for nonviolent felony defendants from the early 2000s. With data on sentencing and risk scores, the research points to three key findings. First, like the findings in Hennepin County, New York City, and in many other studies, Black defendants, when compared to White defendants, are predicted to have a higher risk of court failure or re-offending. Second, the risk instrument directly conflicts with legal principle by assigning high-risk scores to young defendants who are often assessed as less culpable. Third, Virginia judges apply the risk scores in a racially disparate way, being more likely to follow the leniency recommendations of the score for White defendants than for Black defendants.
Fines and Fees Mitigation
Several jurisdictions have either taken an alternate approach to fee collection or pardoned outstanding debts in their entirety in an effort to reduce their disproportionate impact on Black, Latino, and Native American communities. The San Francisco Superior Court was the first in California to stop license suspension as a consequence of failure to pay. The expulsion of the practice is largely due to vocal advocates, who noted that court debt and its consequences fall disproportionately on poor people and Black, Latino, and Native American individuals, further exacerbating cycles of poverty. While no research has been conducted to see if disparity in the system has lessened as a result of this provision, one study found that this practice has had no negative impact on revenue collection (Brown et al., 2020). Another field experiment, conducted in a misdemeanor court in Oklahoma, repaid all current and past court debt for a randomly selected treatment group, and similarly found a very low level of debt repayment in the control group. The Oklahoma study found no racial differences in the effects of clearing court debt, but large racial disparities in the caseloads meant that the fewer warrants, debts in collection, and tax-refund garnishing enjoyed by the treatment group disproportionately benefited people of color (Pager et al., 2022).
Statewide efforts at legal debt relief have been undertaken in Minnesota and Iowa through programming that helps drivers set up a payment plan for outstanding balances and thus avoid license suspension (Bastien, 2017; Schwartztol, 2017). Some jurisdictions are also experimenting with warrant clearances: Atlanta Municipal Court, Milwaukee Municipal Court, and Montgomery Municipal Justice Center all have regular “amnesty periods,” during which those with outstanding minor offenses and traffic violations can resolve cases from the year prior without fear of arrest or of being charged with additional failure-to-appear fees (however, persons are still
responsible for paying prior debt; Bastien, 2017). Lastly, minority-majority districts, like New Orleans Parish, have experimented with bail reform, including mandating ability-to-pay evaluations and consideration of non-monetary bail alternatives (Sledge, 2021).
Modification of Judicial and Prosecutorial Discretion
In 2005, the Vera Institute of Justice created the Prosecution and Racial Justice Program, which aimed to “manage the exercise of discretion within their [prosecutor’s] offices in a manner that reduces the risk of racial disparity in the decision-making process” (Davis, 2013, p. 837). The program partnered with Mecklenberg County (NC) and Milwaukee County (WI) and with their respective district attorney’s offices to identify disparities in prosecutorial decision making and enact policies to reduce them.
Mecklenberg County reduced the number of drug charges they prosecuted by nine percent when they found defendants to be disproportionately Black. When Milwaukee County found that a large percentage of drug paraphernalia cases involved possession of a crack pipe by Black defendants in the city of Milwaukee, District Attorney John Chisholm instructed his office to decline prosecution and instead divert arrestees to drug treatment. Milwaukee also conducted an evaluation of prosecution against domestic violence charges and found that there was a 34 percent greater chance of prosecution when the defendant was Black and the victim was White than when both defendant and victim were White. Chisholm responded with department-wide, culturally conscious training around domestic assault in diverse scenarios (Davis, 2013).
Suffolk County (MA), which includes the Greater Boston Area, has also experimented with changing prosecutorial policy to reduce racial disparity. District Attorney Rachael Rollins, who was inaugurated in 2019, created an officewide presumption of non-prosecution for 15 nonviolent misdemeanor offenses, all of which were considered “crimes of poverty” and disproportionately charged against people of color. The “Rollins list” included disorderly conduct, drug possession, receiving stolen property, and shoplifting (Daniel, 2018). While there has been no evaluation of racial disparity post-implementation at this time, one study found that the presumption of non-prosecution instituted by Rollins’s office was associated with a 47 percent decrease in new criminal complaints for nonviolent misdemeanor cases not on the Rollins list and a 56 percent decrease in new criminal complaints for all nonviolent misdemeanor cases (both statistically significant). The study also found that the new policy had no effect on reported crime (Agan et al., 2021).
Similar to Rollins, other efforts by a newly elected cohort of prosecutors who campaigned to reduce the negative effects of courts and incarceration
have been implemented with the goal of reducing disproportionate system contact for Black, Latino, and Native American individuals.12 While these efforts are race-neutral on the surface, they largely affect Black, Latino, and Native American individuals, either because these jurisdictions have large minority populations or because of racial disparities in court involvement. Additional research is needed to understand the relationship among these reform efforts, racial disparities, and impacts on community safety.
Lastly, one study has identified targeted racial disparities in judicial decision making when sentencing post-conviction. Scorecards that rated the extremity of disparity were calculated at the county level in New York State. These scorecards were made available to both county judicial leadership and the New York State Permanent Commission on Sentencing, but the commission was disbanded before action could be taken based on county-level scoring (Ridgeway et al., 2020).
SENTENCING AND CORRECTIONS
The great increase in correctional populations during the last quarter of the 20th century remains one of the most important and distinguishing characteristics of the U.S. criminal justice system. Over roughly two decades, the United States went from having per-capita correctional populations comparable to those of western European nations to having the world’s highest incarceration rate. While U.S. incarceration rates have declined in recent years, and more so for Black Americans than for other groups, the country still incarcerates its citizens at rates that are high by both historical and international standards, with Black Americans incarcerated at rates many multiples those of other groups. Moreover, the increases in the nation’s prison and jail populations have occurred hand-in-hand with increases in the population of individuals under community corrections. As discussed above, this interacts with policing in a specific manner, since individuals on community corrections are subject to additional scrutiny
12 For example, during her first three years as Cook County (Chicago) state’s attorney, Kim Foxx dropped felony charges against 30 percent of felony defendants, an increase of 10 percentage points compared to her predecessor. Foxx’s office has adopted a range of measures to encourage dropping charges tainted by improper police procedure (Jackson et al., 2020). Prosecutors under Larry Krasner, Philadelphia County district attorney, have reduced requests for cash bail at arraignment (Booker, 2021). George Gascón has dismissed 60,000 marijuana convictions as Los Angeles County district attorney and 9,000 during his former post as San Francisco district attorney, all of which occurred recently but prior to marijuana legalization under CA Proposition 64 (Queally, 2021). Lastly, Brooklyn district attorney Eric Gonzalez established a 2020 action plan that would seal or expunge marijuana convictions and normalize prosecution’s pursuit of non-jail resolutions at all case junctures, among other initiatives (Brooklyn County District Attorney’s Office, 2020). Obstacles and risks associated with changes in prosecutorial policy are examined by Mitchell and colleagues (2022) and Hogan (2022).
and subject to warrantless search. Given racial disparities in rates of incarceration and rates of community corrections supervision, these disparities in community corrections involvement translate directly into disparities in invasive interactions with the police.
To reverse these trends and to narrow disparities in correctional supervision, one could pursue several alternative lines of analysis. One could investigate the specific policy changes that led to large increases in U.S. correctional populations, critically evaluate their impacts on communities in terms of both public safety and harms visited upon those bearing the brunt of the punishment, and then reverse the changes in sentencing policy that have done the most harm and generated the least benefit. Alternatively, one could look to recent experiences with reforms occurring at the federal and state levels, identify those that have had an especially strong impact in reducing correctional populations without affecting public safety, and then devise proposals that draw upon the lessons from these reforms.
The sentencing policy changes in the 1980s and 1990s that drove the increases in correctional populations are well known (see NRC, 2014, Chapter 3). During that period, sentencing policy across the states shifted from having relatively unstructured, indeterminate sentencing schemes in which individuals would receive a wide range of minimum to maximum sentences and correctional authorities would largely determine time of release, to more structured, determinate sentencing with legislatively prescribed sentences and administrative rules pertaining to good conduct credits determining release dates. Sentencing in general became more punitive, both in the widening range of offenses for which prison was prescribed and in the duration of sentences (Tonry, 2019). In many states as well as in the federal system, sentencing guidelines greatly reduced judicial discretion in admission to prison and parole board discretion at the end of a prison term.
Moreover, during this period state and federal lawmakers passed legislation mandating minimum sentences for specific offenses, imposed enhancements for second- and third-time prior convictions or enhancements for specific places or associations (e.g., school zones or gang membership), and added enhancements associated with specific conduct (e.g., use of a gun) or for specific criminal histories (e.g., enhancements for prior prison terms). Many states lengthened mandatory community correctional supervision time and often used technical violations of conditions of release to return people to prison without a new conviction. Moreover, many jurisdictions required that those sent to prison serve minimum percentages of their sentences regardless of conduct while incarcerated or efforts toward rehabilitation. The cumulative impact of these changes lengthened time served, especially for violent offenses; increased prison admission rates, particularly for lower-level drug offenses; increased racial disparities from
the early 1970s to the late 1990s; and largely drove the five-fold increase in the nation’s incarceration rate (NRC, 2014).
Given that U.S. sentencing occurs in the context of 51 different systems, the description above is stylized and the particulars of sentencing (the degree of indeterminacy, the use of enhancements, and the severity of sentences, for example) vary greatly across systems. However, some generalizations can be drawn from the past four decades that may guide reforms to reduce racial inequality. First, part of the growth in prison populations and in racial disparity from the early 1970s to the early 1990s is due to sentencing policies that impose terms of imprisonment and affect the rate of admissions to prisons. Mandatory minimum prison sentences and classification of offenses as felonies (which carry prison time) are examples of policies that tend to increase prison admission rates. Second, part of the growth in imprisonment and racial disparities is related to increases in the duration of sentences. Reforming sentencing policy to reduce racial disparity could thus focus on reducing prison admission rates, or on reducing the length of stay in prison, particularly for those serving long sentences.
Reducing Prison Admissions
Examples of sentencing reforms that have reduced prison admissions include changes in drug policy and “de-felonization” statutes that downgrade felony offenses to misdemeanors (Elderbroom and Durnan, 2018). California’s Proposition 47 provides an important example of de-felonization. Passed in 2014, Prop 47 redefined some nonviolent property crimes, those where the value does not exceed $950, as misdemeanors. It also redefined some simple drug possession offenses as misdemeanors. The law also allows for past convictions for these charges to be reduced to misdemeanors by a court. Figure 8-1 documents arrest-age profiles for the one-year periods prior to and following the passage of Proposition 47. The figure presents felony arrest rates per 100,000 by single year of age for violent offenses, property crime offenses, drug offenses, and other offenses for African American, Hispanic, White, and Asian California residents. It shows that sizable changes occurred for all age categories and that racial disparities narrowed for most age groups for property felony and drug felony arrests. The declines in drug felony arrest rates are very steep and fairly evenly distributed across age groups. The declines in property crime arrest rates are also notable and particularly large for African Americans in their early 20s.
A number of other states have since followed California in reclassifying drug possession from a felony to a misdemeanor, including Utah, Connecticut, Oklahoma, and Alaska. In each case, simple drug possession (until a third conviction) was downgraded from a felony to a misdemeanor, and
people convicted of drug possession could not be sent to prison. Moreover, these changes applied to virtually all controlled substances.
Similar sentencing reforms have been adopted that also reduce prison admission for drug possession, small sales of drugs, and other high-disparity offenses. In particular, reforms to narrow the application of mandatory minimum sentences have been adopted since the early 2000s. Legislative reforms have focused on drug sentencing. A 2014 report found that judges
have been given discretion by state legislatures to depart from mandatory minimum drug sentences in six states: Connecticut (2001), Georgia (2013), Hawaii (2013), Louisiana (2012), Michigan (2002), and New York (2009; Subramanian and Delaney, 2014).
Reducing Long Sentences
As discussed earlier, the growth in prison populations and racial disparities was associated with increasing lengths of stay in prison, especially for violent crimes. The increase in the length of stay resulted from the proliferation of truth-in-sentencing, three-strikes, life-without-parole, and related reforms that concentrated their effects on violent offenses. Since African Americans are incarcerated at higher rates for violent offenses, murder and robbery in particular, the increase in the length of stay has tended to increase the racial disparity in imprisonment (Kazemian, 2022). Neal and Rick (2016), using data from seven states, find that for the violent crimes of murder and robbery, arrest rates declined from 1985 to 2005, yet incarceration in these offense categories increased. This occurred in part because of a slight increase in the risk of prison admission, given an arrest, but mostly because of an increase in the length of stay in prison. The proportion who were in state prison after 10 years increased during the period by 57 percent for murder and by 236 percent for robbery. Similar findings for length of stay were reported by Raphael and Stoll (2013) and Blumstein and Beck (2012; Beck and Blumstein, 2018). In 2020, more than 200,000 people were serving life sentences, and one in five imprisoned Black men were incarcerated on a life sentence (Nellis, 2021).
The increasing length of sentences has tended to prolong the incarceration of many, well past the age of being criminally active. Thus, studies of incapacitation, which estimate the counterfactual criminal involvement in free society of those who are incarcerated, find little to no crime-reducing effect of lengthening already long sentences (NRC, 2014, pp. 140–145). Deterrence studies also find little evidence of crime-reducing effects. Studies of gun crime enhancements (Raphael and Ludwig, 2003), greater penalties associated with moving from the juvenile to the adult justice system (Hjalmarsson, 2009; Lee and McCrary, 2009), and California’s third-strike enhancement indicate that adding time to long sentences either has no effect on crime or, in some studies, only a small effect. Nagin (2013, p. 231) concludes, “The deterrent return to increasing an already long sentence is small, possibly zero.” Largely based on this empirical evidence, the National Research Council panel on high incarceration rates concluded that long sentences could be reduced with very little effect on crime (NRC, 2014, p. 5). Reducing long sentences would also reduce absolute racial disparities because the conviction offenses of Black men disproportionately carry long sentences.
Unlike sentencing reforms that reduce prison admissions for drug possession and other low-level felonies, there are few significant examples of reducing long sentences. Exceptions include California’s three-strikes reform, which greatly reduced the use of long sentences for third-time felonies and eliminated the sentence of mandatory life without parole for people committing homicide as juveniles—which had been ruled unconstitutional by the U.S. Supreme Court in 2012 (Daftary-Kapur et al., 2022). Reducing long sentences in both cases disproportionately benefited Black incarcerated people because of their relatively high incarceration rates, and it had no measurable effect on serious crime.
Another approach to the reduction of long sentences has involved so-called second-look provisions, which offer procedures for review after sentencing. In the Model Penal Code, a second-look review could reduce sentences for good conduct in prison, for advanced age or infirmity, or based on changed circumstances of the offense or the offender after the sentence was imposed (Frase, 2009). Under revisions in 2017, the Model Penal Code recommends legislature authorize judicial review of sentences after 15 years of imprisonment for adult crimes, and after 10 years for youth who commit crimes while younger than age 18. Second-look provisions have been adopted or are being considered in several jurisdictions (Porter, 2021). California’s AB 2942 allows district attorneys to initiate re-sentencing, considering factors such as the applicant’s prison disciplinary record and record of rehabilitation, and whether age, time served, or diminished physical condition has reduced the applicant’s risk of future violence. In Washington (DC) the Second Look Amendment Act, 2020, allows those who committed crime as young adults (under the age of 25) to petition for resentencing after 15 years of imprisonment. Proposed legislation in New York State, known as the Elder Parole Bill, would allow people ages 55 and older who have served 15 years or more to receive a parole hearing. At the federal level, a Second Look Bill has been introduced that would allow people who have been incarcerated for at least 10 years in federal prison to petition a court for re-sentencing.
Although the details of second look provisions vary across jurisdictions, they offer a mechanism to reduce very long prison sentences, and they recognize that the risk of re-offending is relatively low for people in midlife or who are elderly. The provisions are consistent with a large body of research showing very low rates of criminal involvement for people in their thirties, forties, and older (Daftary-Kapur et al., 2022; Blumstein and Nakamura, 2009; Laub and Sampson, 2003; Hirschi and Gottfredson, 1983).
Repealing the Death Penalty
At the far end of sentencing severity lies the death penalty. Chapter 2 reports evidence of both racial disparity and racial bias in death penalty
sentencing, and the committee sees death penalty abolition as an important initiative for reducing racial inequality. The U.S. Supreme Court, in the landmark case Furman v. Georgia,13 invalidated all the states’ death penalty statutes. Furman, which was decided 5–4, was a complex decision with no controlling opinion on which the majority could agree. In deciding with the majority, Justice Douglas spoke directly to the issue of racial inequality and observed that the death penalty was applied disproportionately to “the Negro, and the members of unpopular groups,” noting that in the present cases the defendants were Black and the victims were White (Garland, 2010, p. 226). The sequel to Furman, Gregg v. Georgia,14 reaffirmed the constitutionality of the death penalty with additional procedural safeguards.
Since Gregg, advocacy campaigns against the death penalty have appealed to a variety of rationales, including the incidence of wrongful conviction, the absence of a deterrent effect, the cost of capital cases, the evolving standards of decency, the cruelty of methods of execution, the arbitrariness of capital charging, and what Steiker and Steiker (2016, p. 110) have called the “unjust influence of race in the capital punishment process” (Breyer, 2016; Martin, 2009; see Garland, 2010, Chapters 9 and 10, for a discussion of the politics and culture of the death penalty after Furman).
Racial inequality thus figures as one of several arguments used against the death penalty, although experimental research also shows that presenting evidence of racial disparity can increase support for the death penalty among White people (Peffley and Hurwitz, 2007). In 2005, the U.S. Supreme Court in Simmons v. Roper15 also banned the death penalty for juvenile offenders, citing the neuroscience of adolescent brain development. Thirty-seven states had re-established the death penalty after the Furman decision, but by 2020 repeals had been passed in 12 states, and moratoriums on the use of the penalty had been issued by governors in another three states. Many states, although having a death penalty, have not conducted an execution in years. The Death Penalty Information Center has found that official misconduct is extremely prevalent in death penalty cases—5.6 percent of all death sentences since 1972 have been exonerated due to prosecutorial misconduct. The most likely outcome of an individual’s death sentence is that it will be overturned without being re-imposed. Since 1972, less than one in six death sentences have resulted in an execution (Death Penalty Information Center, 2022b). In 2020, the federal government carried out 10 of the 17 executions nationwide.
13 408 U.S. 238 (1972). See https://supreme.justia.com/cases/federal/us/408/238/
14 428 U.S. 153 (1976). See https://supreme.justia.com/cases/federal/us/428/153/
15 543 U.S. 551 (2005). See https://supreme.justia.com/cases/federal/us/543/551/
Probation and parole are systems of supervision that hinge on a high degree of discretion. Entry into probation and revocation for technical violations are both determined by a judge in a court hearing. Parole supervision also operates with a high level of discretion, such that a parole board first decides on early release from prison and later on return custody in the event of a violation. Research shows large racial disparities in the rates of both community supervision and revocation of community supervision, with Black and Hispanic people at greater risk of re-incarceration (Bradner et al., 2020; Steen et al., 2013; Grattet et al., 2009). Black parolees and probationers are regularly found to be at high risk of return to custody for violations of technical conditions of supervision and for new offenses (Bradner and Schiraldi, 2020; Lin et al., 2010; Steen and Opsal, 2007). The conditions of supervision are themselves often onerous, involving regular visits with probation and parole officers, drug testing, mandated programming, unannounced visits by officers to places of work and home, and the payment of supervision fees.
Recent reforms have reduced the disproportionate impact on minority parolees and probationers by limiting the duration of community supervision, limiting discretion to revoke parole or probation, and reducing the intensity of supervision. Reforms in all these areas appear to have had very little, if any, adverse effect on crime rates based on available research.
The largest recent reduction in the scale of community supervision can be found in California, as part of the process of realignment, initiated by a series of cases pointing to constitutional violations as a result of prison overcrowding. The court rulings and advocacy prompted legislation that reduced the scale of imprisonment and the extent of community corrections supervision. In 2009, the California legislature passed Senate Bill 678, the Community Corrections Performance Incentive Act,16 which created incentives for counties to retain probation supervision for technical violations and eliminated parole for low-level offenses. In 2011, the California Public Safety Realignment Bill legislated that felony convictions for nonserious, nonviolent, and nonsexual offenses could no longer result in sentencing to state prison and parole but only to local jail and probation. Restricting community supervision in this way also meant that probation revocation would result in 90 days in jail and not in longer periods of imprisonment (Sundt et al., 2016). In 2014, California’s Safe Neighborhoods and Schools Act (Proposition 47)17 downgraded six felony offenses to misdemeanors, reducing the application of prison sentences (and subsequent parole) and
16 See https://www.courts.ca.gov/documents/sb678.pdf
felony probation, and diverting people to shorter and often unsupervised periods of local probation. In 2020, Assembly Bill 195018 was passed, reducing misdemeanor probation in California from a maximum of three years to one, and reducing felony probation from five years to two. In the period from 2009 to 2020, California experienced steep declines in the probation revocation rate, which dropped from eight percent to 2.5 percent.
The California reforms, which were sustained over a decade, included a mix of sentencing reforms and administrative changes that diverted jurisdiction from the state to the local level. In many cases, reducing the severity of sentences—sometimes through sentence reductions and sometimes through de-felonization—had the effect of eliminating parole supervision or of shifting people from state felony probation to the less-punitive local misdemeanor probation.
A number of studies have examined the impact of these California reforms on the state’s crime rates. During the period of declining community supervision in the 2010s, crime declined by 7.4 percent (Bartos and Kubrin, 2018), and “no measurable effect on violent crime” from the California reforms was found. Analyses of California Realignment and Proposition 47 similarly found no increase in violent crime and small declines in re-arrest and reconviction (Bird et al., 2018). While these analyses do not isolate the effect of reduced community supervision, they indicate that the overall scale of correctional supervision—institutional and community-based—has been reduced significantly with no adverse effect on serious crime.
A second case study of reducing the scale of community supervision is provided by New York City (Lopoo et al., 2022). In New York City, pretrial populations can be assigned to probation supervision, and misdemeanor convictions can result in a sentence of local probation. Beginning in the early 1990s, New York City greatly reduced the intensity of probation supervision, using reporting kiosks that followed a fingerprint reading with a few questions instead of the traditional meeting with a probation officer. The intensity of supervision was further reduced in the 2000s with “distance reporting,” in which probationers could check in by phone or computer. Probation officers would request a reduction in the period of supervision for those completing distance reporting successfully. New York City also undertook a large-scale review of probation absconders, reviewing more than 15,000 records; if an absconding person had not been re-arrested, the arrest warrant for absconding was dismissed. Policy on probation revocation also shifted in the 2000s to discourage incarceration for violations of the conditions of supervision.
The cumulative effect of policy reforms in New York City, along with substantial reductions in felony arrests from the early 1990s to the 2000s,
18 See https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=201920200AB1950
resulted in a large decline in local probation supervision. In 2000, probation supervision in New York City peaked at 82,342. By 2021, the city probation population had shrunk to 11,531. The index crime rate fell as well, from 3,099 per 100,000 (in 2000) to 1,732 per 100,000 (in 2019) during the period of shrinking community supervision (Lopoo et al., 2022). The decline in the probation population also unfolded in a context in which advocacy groups raised public awareness about criminal justice reform, and an array of social services for job training, housing, educational assistance, and peer mentorship were made available to those involved in the criminal courts (Lopoo et al., 2022). These trends in New York City probation supervision, along with other research finding no crime-reducing effect of probation or parole, suggest that substantial reductions in community supervision are possible without adverse effects on public safety.
In sum, the experiences of California and New York City indicate large reductions in the probation and parole population, with substantial and disproportionate reductions in the community supervision of Black and Hispanic people. These reductions were achieved with a wide variety of policy changes, including reforms to sentencing those downgraded offenses from felonies to misdemeanors, that limited parole and probation periods and that shifted jurisdiction for supervision from the state to the local level. In addition to legislative shifts, policy change within probation departments has reduced the intensity of supervision, shifting from personal reporting to an officer to various forms of distance reporting. All these reforms were adopted during periods of significant reductions in crime.
These experiences, sustained in very large jurisdictions over several decades, indicate that substantial and disproportionate minority correctional supervision can be reduced through sentencing reforms that limit the numbers and duration of probation and parole supervision and that reduce revocation to incarceration for technical violations, as well as through agency policy changes that reduce the intensity of supervision.
The steady accumulation of sentencing reforms around the country since the early 2000s has contributed to reductions both in the prison population and in racial disparities. Changes in drug sentencing, notably the elimination of mandatory prison sentences, have reduced incarceration for drug crimes, reducing absolute racial inequality between Black and White people. Reducing long sentences also promises to reduce racial disparity, with Second Look provisions offering one important mechanism for reviewing long sentences. Several states have also reduced community supervision populations and curtailed the discretionary power of line officers and judges to return people on probation and parole to incarceration for technical violations. Reducing revocations for technical violations of community supervision is also likely to reduce racial disparities in incarceration.
The U.S. constitutional system embodies a principle of parsimony, of limited government, that is reflected in foundational restrictions on executive powers of search and seizure, the prohibition of cruel and unusual punishment, a positive right to counsel, and due process protections against group-based inequalities. In the era of mass criminalization and high incarceration rates, the principle of parsimony was subject to enormous stress. Many of the criminal justice reforms reviewed in this chapter restore the principle of parsimony by reducing police contact, the scale of incarceration, and arbitrary community supervision. In so doing, these measures also offer substantial promise for reducing racial inequality.
Although racial inequality in criminal justice contact remains high, a wide range of criminal justice reforms have been adopted across the country that have reduced the disproportionate involvement in disadvantaged communities. Reforms—variously imposed by courts, by federal monitors, through legislation, or through agency policy change—have changed the practice of policing, court processing, sentencing, and correctional supervision. Clear evidence of reduced racial inequality is associated with large reductions in the scale of police stops, the level of incarceration, and revocations of parole and probation supervision. In each of these cases, we see evidence of large reductions in absolute racial disparity at little or no cost to public safety.
Efforts at reducing relative disparity—whether with body-worn cameras, risk assessments, or anti-bias training—are not as strongly supported by the research evidence. This is in part because the evidence base is relatively thin, and in part because technical and human interventions may be underpowered in the face of the large structural inequalities in which the criminal justice system is embedded.
Throughout this review we have encountered the complex issue of possible tradeoffs between racial fairness and crime. Does reducing racial inequality come at the cost of more crime? In many cases, fairness is constitutionally mandated and in this sense is a dominant value for public policy. In some cases, the harms of criminal justice intervention are clear-cut, and reducing the scale of intervention both reduces racial inequality and promotes safety. The criminogenic effects and pains of long periods of incarceration is a leading example. But our review also points to examples, such as directed police patrols, where the goals of crime reduction and racial fairness—at least in the very short term—can sometimes be in conflict. In these cases, the risk of greater inequality can be addressed by designing criminal justice interventions (say, gun patrols) with strong safeguards against abuse and using them in a limited way in coordination with close community collaboration and social policy.
CONCLUSION 8-1: A wide variety of measures—including judicial bans on unconstitutional policing and incarceration, sentencing reform for drug offenses and de-felonization, bail reform, and reductions in the intensity and duration of community corrections supervision—have reduced the overall level of criminal justice contact, incarceration, and community supervision. Although in many cases relative disparities have not been reduced, these measures have had large effects on reducing absolute racial disparities, with little evidence across specific cases of an adverse effect on crime.
CONCLUSION 8-2: More targeted initiatives to reduce relative disparities—through, for example, federal oversight, diversity hiring, anti-bias training, or quantitative risk assessment—have been less clearly effective. In part, this is due to limitations of the research evidence, which should be further developed to effectively guide policy. Targeted initiatives also face resistance from criminal justice officials—police officers, police executives, and judges, for example—who may not view the initiatives as legitimate.
Although some jurisdictions have significantly reduced specific kinds of criminal justice contacts that have in turn reduced absolute disparities, substantial racial inequality in the criminal justice system remains.
RECOMMENDATION 8-1: Subject to the main goals of parsimony and community safety, states, localities, and the federal government should explore ways to reduce police stops and searches, jail detention, prison admissions, and long sentences, which would further reduce racial disparities. Examples of such efforts could include limiting jail detention to only those charged with serious crimes who pose a serious and immediate risk of harm or flight, pursuing further drug sentencing reform, establishing second-look provisions for long sentences, eliminating revocations of community supervision for technical violations, and eliminating the death penalty.
CONCLUSION 8-3: The research and policy experience reviewed here has often involved piecemeal and uncoordinated reforms that have affected racial disparity at specific stages of criminal processing but have had less effect on racial inequality that emerges across stages and in community contexts of racial inequality. Reducing racial inequality will involve coordinated reforms across stages of the criminal justice system that will reduce the racial disadvantage that accumulates from police contact, to court processing and sentencing, to correctional supervision.
RECOMMENDATION 8-2: Given evidence of how racial inequality in the criminal justice system partly results from structural inequalities in society and the cumulative effect of criminal processing, states, localities, and the federal government should explore ways to reduce racial inequality through coordinated reforms that work across the stages of the criminal justice system and also address structural inequalities in society.
This page intentionally left blank.