Frontiers of Research on Racial Inequalities in Criminal Justice

UNIFYING THEMES

The first theme I highlight is the need to foreground social change and its community contexts of racial inequalities in criminal justice. As Glen H. Elder Jr. and his colleagues write, “The life course of individuals is embedded in and shaped by the historical times and places they experience over their lifetime” (Elder et al. 2003, 12). Like the society in which it operates, the apparatus of criminal justice in the US is not static or monolithic—its many variations over time and place require explanation (Sampson, forthcoming).

To take a leading challenge this issue explores, if racial inequities are seen as inherent to the operation of American punishment, how do we explain the large variations in rates of incarceration of racial groups across regions, states, and communities (National Research Council 2014; National Academies of Sciences 2023) or the substantial declines in overall incarceration and in Black-White disparities in incarceration in the past fifteen years (Robey et al. 2023; Muller and Roehrkasse 2025)? Are we on the road to ending mass incarceration, and if so, how does this bear on accounts that view it as a new slavery or Jim Crow? Or has the form of punishment instead changed, an argument that follows from Sara Wakefield and Kristin Turney’s (2025) review in this issue on the rise of the carceral state, which they define to include other features of the criminal legal system (such as community supervision and legal financial obligations) and surveillance from institutions adjacent to the criminal legal system (such as schools, child welfare agencies, and the health care system).

Either way, we need conceptual and empirical frameworks for understanding contextual variations and sociohistorical change, especially profound cohort inequalities in criminal justice contact and punishment—what Roland Neil and Robert Sampson (2021) term the “birth lottery of history.” Even within relatively short periods, social change reconfigured the incarceration chances of successive birth cohorts in racially disparate ways (Pettit and Western 2004), most recently in surprising directions. A focus on social change and the contexts that mediate it centers not just cohort differentiation but also changing racial heterogeneity in the collateral consequences of criminal justice contact (Lee et al. 2025, this issue), along with the ways in which definitions of crime and perceptions of criminals have changed. The evolving nature of how immigrant groups in America have been conceptualized and treated in the criminal justice system is a case in point (Muhammad 2011a, 2011b), foreshadowing contemporary concerns with what some call “crimmigration” (Ryo et al. 2025, this issue). Each article in this issue is attuned in its own fashion to the changing nature of criminal justice and its societal context.

But crime has changed dramatically over time too, and hence the second theme that I argue is central to theoretical advancement is uniting the study of crime and criminal justice. Past research has tended to split into two camps, one highlighting the study of criminal behavior and the second the study of criminal legal contact and formal punishment (Laub 2014). Especially during the era of mass incarceration, these two frameworks became increasingly separated, perhaps best captured in Michelle Alexander’s (2012, 215) argument that since “all of us are criminals,” criminal behavior does not much matter, if at all, in understanding punitive crime policies. The crack cocaine legislation that President Ronald Reagan signed into law in the 1980s is a leading example of how behavior can become untethered from punitive policies that are racially discriminatory. Yet as James Forman (2012, 2017) countered, and more recently David Garland (2025), to ignore the reality of changes in crime and violence, especially in poor minority communities, is to miss an important part of the story of when, where, and why incarceration rose as it did. Research has also shown that self-reported criminal behavior and victimization predict official criminal justice indicators, such as arrest, and that there is broad consistency in the social class and race correlates of official and self-reported data for similar offenses (Hindelang et al. 1979; Elliott and Ageton 1980), though these connections may have weakened over time at the individual level as incarceration increased (Weaver et al. 2019).

It follows that a key question, one the articles in this issue may be put in dialogue with, is: under what conditions does crime, including changes in crime, matter for our understanding of racial inequalities in criminal justice contact and beyond, and how? Rather than take an either-or position, this question cuts across time and across levels of analysis to include individuals, communities, and entire societies. Violent victimization in particular is a devastating experience that disproportionately harms Black Americans, undermining entire communities and leading to lifelong consequences for children exposed to violence (National Academies of Sciences, Engineering, and Medicine 2023, chap. 3). Hence the more indirect the collateral consequences of criminal justice contact we start to consider (whether in families, extended families, or communities), the more important it is to simultaneously know how crime and violence matter. Not only does integrating the study of criminal behavior and criminal justice widen our view of racial inequalities (Laub 2014), but it also sharpens debates on what it means to say that the US is a penal outlier. Some have argued that relative to its levels of crime, the US is not as exceptional in its punishment levels as is commonly asserted (Clegg et al. 2024).

A third and interrelated theme is that the articles in this issue focus largely on structural causes of racial disparities, especially in ways that recognize the sequential nature of criminal justice processing and life course dynamics more generally. John Dovidio and Phillip Atiba Solomon (2025, this issue), for example, distinguish the implicit bias of individual police officers from more systemic or supra-individual sources of bias like racial segregation, or more generally what some call structural racism in criminal justice institutions (National Academies of Sciences, Engineering, and Medicine 2023, 22; Lee 2024). Shawn Bushway and colleagues (2025, this issue) also stress structural racism, examining the form of punitive laws and practices that create disparate impacts for minority defendants without generating improvements in public safety, identifying a key mechanism that explains racial disparities in prison sentences and that, like Dovidio and Solomon’s (2025) assessment, does not require individual bias on the part of criminal justice decision-makers.

Another contribution—by Alia Nahra and colleagues (2025, this issue)—takes a more explicit life course perspective in confronting the challenge of individuals returning to the community after a prison sentence, highlighting the problems of compounded disadvantages that accumulate over time and that disproportionately impact poor Black men in their efforts to reintegrate back into society (Miller 2021; Western 2015). It is not just the criminal justice system that is in play, but also social welfare, health, education, and many other features of racial stratification in America. Relatedly, in their comprehensive survey, Hedwig Lee and colleagues (2025) raise hard questions of causality and selection bias in research on the consequences of contact with legal authorities at different stages of the life course, noting heterogeneous effects by race. They also highlight the role of intergenerational criminal justice contact in fostering racial inequality, emphasizing the concept of linked lives (Elder 1994; Wildeman 2020).

Taken together, this issue’s articles stress the importance of structural accounts of racial inequalities that extend well beyond individual-level bias and policies purporting to eliminate individual prejudice by criminal justice agents. Coupled with a focus on the fundamental problem of violence in America (Zimring and Hawkins 1999) and the changing nature of both crime and punishment over the past quarter century, a new generation of research is in order.

THE SOCIAL CHANGE CHALLENGE

Much has been written about incarceration’s dramatic rise starting in the mid-1970s and its collateral consequences for racial inequality, as the articles in this issue attest. Less is known about both earlier and more recent points in history, but new research is expanding its reach both back in time and forward to recent developments in penal practices, raising a new set of questions.

One welcome direction is a push for more historically based research that aims to link distal and proximate mechanisms of incarceration (Hinton and Cook 2021). Instead of just focusing on contemporaneous causes of racial disparities, in other words, scholars have examined the social processes leading up to the penal changes that began in the 1970s. For example, although there was an undeniable and nearly continuous rise in crime and violence in the 1960s, it yielded almost no response in terms of incarceration for about fifteen years. As Forman (2012) and others have discussed, well before the ascent of mass incarceration, Black citizens were vocal in demanding a broad range of societal responses to rising crime and violence, including a greater emphasis on resources such as adequate housing, jobs, education, and more. But the response, as we now know, was lopsided in favor of aggressive policing and, in time, prison. Tracing the “road not taken”—despite explicit warnings, notably by the National Advisory Commission on Civil Disorders (1968), that emphasized the social foundations of racial inequality and the dangers of purely criminal punishment approaches—is a needed corrective (Hinton 2016).

As Nicola Lacey and colleagues (2018, 119) further point out, even though the proximate driver of incarceration lies in the policy choices of legislators, how those choices are shaped by crime rates and perceptions of crime and social order, which evolve over time, is an important question. They argue, for example, that we should expect a significant time lag between changes in crime and imprisonment practices given the reality of how the American criminal justice system works, especially with its long prison sentences. In this case, it is not just the level of crime that matters but also the rate of change in crime and change in imprisonment, which means that researchers need to be more precise about lag effects and long-term processes of mediation, including public attitudes (see also Enns 2016). Taking this sort of temporal perspective invites a series of questions, such as: Why did sentencing policies change when they did? How much is attributable to changes in crime or perceptions of crime, and how long did it take for sentencing practices to react to these changes? We often hear that crime “per se” does not matter in explaining punishment, but this merely deflects the question of potential indirect pathways and moderating or exacerbating influences.

A related issue turns on the interaction of crime and perceptions of threat with demographic and other structural changes in society. From this perspective, some argue that what mediated a penal rather than social welfare response to increasing crime was the changing meaning assigned to Black people accused of crimes after the civil rights movement, the riots of the mid-late 1960s, and the increasing racial diversity of society. Consistent with this view, Scott Duxbury (2023, 561) reports that criminal threat narratives—media discourse that portrays racial minorities as criminally threatening—accounted for about half of homicide’s effect on the overall incarceration rate. Crucially, Duxbury also found that homicide had an indirect effect on racial disparities in prison admissions through the pathway of criminal threat narratives. Black criminality did not register in the same way as, say, White criminality did in earlier eras among Irish or Italian immigrants (Muhammad 2011b). Punishment, rather than mercy, became the dominant force.

A growing empirical literature also looks at racial disparities in imprisonment that existed well before the late twentieth century, vital though that era was to our understanding of mass incarceration. In addition to work on how European immigrant groups were incorporated into the American mainstream, Christopher Muller (2012) examines the late nineteenth and early twentieth century, connecting in part rising racial disparity in incarceration to increasing African American migration from the South to the North between 1880 and 1950. A new line of work goes back even earlier, examining the incarceration of African Americans from the days of slavery to the present (Muller 2021; Duxbury 2024). This research explores how slavery and later forms of racial domination shaped experiences with incarceration in urban versus rural Black communities that produced racial differences in family structure and employment (Jaynes 2023) that in turn bore downstream consequences for criminal justice experiences. Simply put, for Muller (2021) there was exploitation (relatively low incarceration) in the rural areas and exclusion (relatively high incarceration) in the urban ones. As I will discuss further, the study of mechanisms linking historical inequalities, cumulative early-life disadvantages, and later-life contexts of racial disparities in criminal justice contact is ripe for future inquiry. But for now, the main point is that while the rise in mass incarceration may have been novel in its scope, racial disparities are not, and arguably the causal mechanisms producing them have roots that can be traced back decades if not centuries.

At the other end of the historical spectrum, more attention is being paid to social changes in incarceration and other aspects of criminal legal processing that have ensued in the past two decades. Here, a new set of dynamics seem to be occurring, with race and class inequalities raising several intriguing questions. As Lee and colleagues (2025) document, and other evidence confirms (Muller and Roehrkasse 2022), declines in incarceration in the last twenty years are substantial, and increasingly, class inequality in imprisonment is outpacing racial inequality. Unexpectedly, for example, while the prison admission rate among Black people without any college education has fallen since 2000, the prison admission rate among White people without any college education has steadily increased (Muller and Roehrkasse 2025).³ Using cohort data, Jason Robey and colleagues (2023, 977) similarly show large declines in Black male incarceration relative to Whites, and they also find that, in contrast with just ten years earlier, “Black men are more likely to graduate college than go to prison.”⁴ Meanwhile, and rather remarkably, Black and White disparities in federal court sentences for drug offenses, the trigger for leading accounts of mass incarceration, have disappeared altogether (Light 2022).

It is not just incarceration that has been transformed. Arrest is the main gateway into the criminal justice system, and arrest rates have dropped dramatically too (Lee et al. 2025). In one study from Chicago, the odds of being arrested during the peak ages of delinquency were approximately double for those born in the early to mid-1980s compared to otherwise similar adolescents born just a decade later, in the mid-1990s (Neil and Sampson 2021). Arrests for drugs and disorderly conduct, which are frequently posited as driving mass incarceration, were also substantially lower for the younger cohorts. The authors also show that these changes were not explained by the differences between birth cohorts in demographic composition, family background, economic status, or even early life neighborhood environments (1158). Rather, they found that substantial cohort differences in being arrested in late adolescence and early adulthood arose from the divergent social environments through which each cohort came of age, with Black youth benefitting the most from sharply lower arrest rates in the 2000s.

Given these large declines and changing racial disparities, how much cohort variation in arrest and incarceration is due to changes in penal practices and how much is due to changes in criminal behavior is a question that needs to be addressed. Neil and Sampson (2021) estimate that for arrest, it is about an even split between changes in police practices, such as the collapse of “broken-windows” policing of disorder and drugs, joined with broader societal drops in crime due to behavioral and normative shifts associated with the great American crime decline (Zimring 2006). But the jury is still out on a precise accounting of changes in arrest rates.

Research is also needed to disentangle the policy mechanisms underlying recent changes in incarceration rates. As Bushway and colleagues (2025) argue, policy changes can be decisive independent of variations in criminal behavior. Their analysis suggests that several reforms help explain reductions in incarceration and its associated racial disparities: limiting the use of incarceration for technical violations of parole or probation rules (revocations), introducing alternatives to cash bail, and reforming sentencing through measures such as eliminating three-strikes laws. How these and other policy reforms interact with broader structural inequalities to either mitigate or exacerbate racial gaps remains a key question. For example, under what conditions and at what time horizons do crime and policy matter, to what extent, and for which racial groups? Another question is how much the politics of criminal justice reform drove specific policy changes (Beckett and Brydolf-Horwitz 2020), and whether lower crime provided the context in which reform became possible. Such questions align with a research agenda seeking to link distal and proximate causes of racial disparities.

In short, racial disparities in criminal legal processing are highly variable over time, and both arrest and imprisonment rates have declined substantially in directions that were not predicted by the literature on mass incarceration. Crime and victimization rates have changed markedly too in unexpected ways (Baumer et al. 2018; Sharkey and Marsteller 2022), with consequences for racial disparities in criminal justice. This includes the possibility that the large crime declines of recent decades have had direct effects on reducing racial disparities in criminal justice contact independent of policy changes. Along with a growing scrutiny of diverging class and race-specific trends in intergenerational mobility (Chetty et al. 2024) and what has been termed “deaths of despair” (Case and Deaton 2021), future research will need to resolve important questions on the causes of changing racial inequalities in life chances over the last quarter century.⁵ Cohort is an important analytic category that provides leverage in assessing the nature and mechanisms of these myriad kinds of social changes (Pettit and Western 2004; Shen et al. 2020; Robey et al. 2023; Sampson, forthcoming).⁶

INSTITUTIONAL DIFFUSION AND THE CHANGING CARCERAL STATE?

Although incarceration and arrests have clearly declined in recent decades, it may be that other forms of punishment have taken their place. This issue directly considers this possibility. For Wakefield and Turney (2025), an advantage of the concept of the carceral state is that it emphasizes forms of surveillance and punishment that go beyond incarceration, raising the specter of a changing network of interconnected control institutions. As they note, carceral state scholars argue that the US is more than simply a state with a high crime or punishment rate; rather, it is characterized by an unremitting commitment to seeking carceral solutions to a vast array of social problems. Their viewpoint encompasses features of the criminal legal system, such as police contact, community supervision, and legal financial obligations, in addition to surveillance from institutions adjacent to the criminal legal system, including schools, child welfare agencies, the labor market, the health care system, and immigration control.

Despite the conceptual slippage across different definitions of the carceral state (Rubin and Phelps 2017), and the apparent circular reasoning in asserting that punishment (the outcome) fulfills a societal function that, in turn, contributes to the carceral state’s stability (defined in terms of punishment), the carceral state concept raises empirical questions about mission creep, or the institutional spillover of “carceral logics.” As in classic deviance theory (Cohen 1985), this framing invites us to look beyond incarceration, not just to earlier points in criminal processing like arrest but also to the hypothesized expansion of surveillance and control in adjacent institutions. Kelley Fong (2023) takes this route, studying how child protective services, ostensibly meant to help and protect children in need, may exacerbate inequality through the surveillance and, arguably, punishment of mothers, especially of color. Another conduit is the circulation of people between the health care and criminal justice systems. In Redistributing the Poor, Armando Lara-Millán (2021) argues that underinvestment in health and overinvestment in criminal justice reveal a persistent crisis between budgetary limits and new legal rules that further marginalize poor people of color.

In another kind of spillover, Sarah Brayne (2014) argues that criminal legal contact (being arrested, convicted, or incarcerated) leads to a pulling back from involvement in medical, financial, labor market, and educational institutions that have a surveillance function, what she calls “system avoidance,” a process that severs already marginalized subpopulations, like African Americans, from institutions that support desistance from crime. In the context of immigration, Asad L. Asad (2023, 24) nuances the system avoidance thesis by suggesting that “undocumented immigrants do not evade institutions wholesale; rather, they avoid specific institutional interactions,” a phenomenon he refers to as “selective engagement.”

More research of this type is needed, but what also seems needed is a tighter theoretical definition, consistent across authors, of the carceral state, and thus what constitutes a change in the carceral state. Ex ante predictions are rendered impossible if the scope conditions themselves keep changing or expanding after the fact. In addition, even if we had consistent definitions, the systematic data necessary to assess social change trends and cohort differentiation in racial and ethnic disparities are not widely available in the same way that they are for traditional criminal justice indicators. It seems obvious in retrospect, but we cannot study change or the diffusion of penal control unless we have consistent measures of the phenomenon in question over time and at multiple levels of analysis. Thus, while we may have adequate measures of, say, arrest and incarceration by race over time among individuals, communities, states, and sometimes even countries (Lee et al. 2025), changes in surveillance by institutions beyond the criminal justice system are difficult to address directly given the lack of systematic definitions and long-term data. Filling these theoretical and empirical gaps presents a challenge for the future.

The immigration system in particular is an institution adjacent to the criminal justice system that is in a state of flux, making it a pivotal agenda going forward. Emily Ryo and colleagues (2025) review evidence that the racial and ethnic makeup of local communities shape state and local political decisions on whether and how to engage with federal immigration enforcement efforts. While immigration has long been shown to be unrelated or even negatively related to crime and incarceration (Ousey and Kubrin 2018; Rumbaut and Ewing 2007), perceptions run in the opposite direction and will likely drive even further sentiment for the criminalization of immigrants. Indeed, even as the evidence mounts that immigration is not a major cause of crime (and may even lower it), the evidence that people react punitively to increases in immigration because of perceived crime threats seems to get stronger (Fink and Brady 2019). The externalization of racial threat due to migration flows more globally, along with the heightened visibility that the refugee migrant crisis has played in contemporary politics and policies, suggests that racial disparities in immigration enforcement will be increasingly salient. The disconnect between research and policy in this area is a major challenge in an era of increasing global migration.⁷

Another challenge that Ryo and colleagues (2025) highlight concerns the changing tools of immigration enforcement. They review the rapid growth of immigration enforcement databases, surveillance technologies, and algorithmic risk assessment tools, including the use of artificial intelligence. They describe the increasing linkage between criminal law enforcement and immigration enforcement using these tools, leading to “crimmigration” (see also Menjívar et al. 2018). The turn to predictive technologies is far-reaching and goes well beyond changes in immigration enforcement.

ALGORITHMIC BIAS AND THE PREDICTION CHALLENGE

Prediction has long been a mainstay of both research and policy in the criminal justice system, but the stakes have been raised with changes in the availability of big data, artificial intelligence, and powerful statistical techniques. Consistent with the carceral state argument, for example, police tactics have expanded to new forms in our digital age. Brayne (2020) argues that data-intensive surveillance practices by the police have deepened long-standing mechanisms of social and racial stratification. The worry is that the use of big data and algorithms by police and other legal authorities widens the reach of the criminal justice system, reinforcing racial stereotypes and discrimination.

Predictive risk-assessment instruments (RAIs) have come under particular scrutiny, especially those using machine learning algorithms on administrative data (see, for example, Berk 2018; Ludwig and Mullainathan 2021; Obermeyer et al. 2019). For present purposes, the relevant concern is the possibility that algorithmic risk assessment instruments typically used in the criminal justice system may exhibit racial bias. There is debate over how extensive the problem is (Angwin et al. 2016; Ludwig and Mullainathan 2021), but the research base is relatively small and for the most part lacks definitive tests. Moreover, most research to date is concerned with whether algorithms themselves are racially biased, rather than biases in the data that power them.

But as Roland Neil and Michael Zanger-Tishler (2025) observe in their recent review of an expanding literature, a distinct and underappreciated problem is that most research testing for racial bias in RAIs uses arrest data under the assumption that they constitute an acceptable proxy measure of criminal offending. The key issue is that while a risk assessment might predict future arrest, what we care most about is whether the person being evaluated does or does not commit a future crime. The assumption of a one-to-one connection of offending to arrest goes against a long tradition of research in criminology that instead treats this relationship as problematic and one to be assessed rather than assumed, especially with respect to race and social class (Elliott and Ageton 1980). To the extent that bias exists in arrest, which is used as the outcome of interest, then “label bias,” a phrase used in the prediction literature to refer to measurement error in the outcome, becomes a potentially important problem. Because arrest data are frequently plugged into RAIs, racial differences in arrest as a proxy for offending may be a likely pathway through which RAIs become racially biased. This is not a bias of the algorithm itself but a bias of the data typically used to train and implement the RAIs in practice.

Neil and Zanger-Tishler (2025) assemble multiple strands of evidence, including from outside traditional prediction research, to evaluate racial differences in arrest as a measure of racial differences in crime. They conclude that aggregate-level studies cannot reliably estimate the extent of the problem in the use of predictive algorithms. Perhaps surprisingly, better designed studies at a more disaggregated or micro level come to opposite conclusions: crime incident studies using the National Incident-Based Reporting System tend to find bias against White people, while individual-level self-report studies tend to find bias against Black people. Either way, the implication from incident and individual-level research is that biased arrest records can lead to label bias, thus producing algorithmic bias. Racial bias can also arise, though less consequentially, from feature bias (measurement error in predictors), and sample (or selection) bias in how and on whom the underlying data were collected. Unfortunately, because of these various data limitations, referred to as “dirty data” by the authors and previous scholars (Richardson et al. 2019), existing research is largely uninformative on just how much racial bias exists in common predictive algorithms.

To advance the field, Neil and Zanger-Tishler (2025) recommend a suite of potential solutions that have yet to be systematically explored and applied, such as sensitivity analyses of hypothetical levels of racialized label bias in arrest records and direct comparisons of models that incorporate more accurate offending data, which touches on a long-standing challenge in criminology. For instance, future work might inform arrest-based prediction methods by investigating alternative data sources such as victimization surveys, self-reported data on criminal behavior, or direct behavioral indicators, such as rideshare data on speeding infractions.⁸ The problem is that measurement error exists in all these sources and thus we will only glean an indirect indication of whether algorithmic bias might be a problem, especially since behavioral studies typically are not designed to be used in predictive risk assessments. Information on the links between behavior and arrest could nonetheless still be useful for sensitivity tests that explore how biased available measures would need to be to significantly alter predictions from RAIs. Because predictive tools and the use of RAIs on administrative data have exploded in use, this is a frontier area of research, one that again shows the importance of uniting theories and data on both crime (offending) and actions taken in the criminal legal system.⁹

PENAL EXCEPTIONALISM, COMPARED TO WHAT?

Any consideration of race, criminal behavior, and punishment runs headlong into another major debate—the extent of America’s penchant for punitiveness. The US is by most accounts a penal outlier, racked by incarceration rates well above those of European and other common law countries (National Research Council 2014, 36; Tonry 1995). As a result, the rise of mass incarceration and the extent of its racial disproportionality are often thought to be distinctly American.

But two questions may be raised in this regard. One, do we compare incarceration (or other forms of punishment) relative to a country’s population or to its level of crime, and if the latter, what type of crime? Second, what is the right comparison group to the US in cross-national research on racial disparities in criminal punishment? Analogous to assessing individual-level racial bias in criminal justice contacts, the notion of what “similarly situated” means looms large in both cases.¹⁰ Integrating patterns of crime and punishment permits leverage in assessing America’s penal exceptionalism.

On the first question, scholars such as Marc Mauer (2017), Tapio Lappi-Seppälä (2017), John Clegg and colleagues (2024), and Garland (2025) argue that the punitiveness of the US needs to be measured in conjunction with its levels of crime and disorder, especially violence. When we do so, by some measures the US is less an outlier than is commonly understood in the mass incarceration literature (Wakefield and Turney 2025). In terms of measuring prisoners per homicide, for example, which Clegg and colleagues (2024) argue is a reliable proxy for violence and the type of crime that elicits great social concern, the US looks similar to many European countries, albeit at higher levels of both violence and incarceration. They add that even if we stick to raw rates of imprisonment, the US no longer claims the lead—El Salvador, Rwanda, and Cuba now have the highest rates of imprisonment in the world. Combined with the fact that relative racial disparities in incarceration in the US are similar to those in Canada, Australia, and the UK, Garland (2020, 325) concludes that the US is “not particularly distinctive.”¹¹

The second and related question turns on what the right comparison country is in determining penal exceptionalism. In the existing literature there is heavy reliance on what are typically called most-similar-systems research designs, whereby analysts compare countries with allegedly similar characteristics but different outcomes, in this case punishment (Arruda de Amaral 2024). Here we get the typical Canadian and especially European comparisons to the US. But another approach switches things around by considering countries with similar outcomes, in this case high punishment (and perhaps crime), yet different socioeconomic or political characteristics (most-different-systems design). By this logic, we might look more to the Global South and countries such as Brazil rather than Europe when seeking comparisons for America (Parker and Perez-Brumer 2024). Lacey and colleagues (2018, 212) likewise argue that we should be looking at “how penal policy is determined in emerging or transitional economies such as those of South America, South Africa, and Eastern Europe, or in the mixed market economies of Southern Europe.” Intriguing historical questions arise from this vantage point. For example, might the legacy of Jim Crow for racial disparity in punishment share a mechanism with that of apartheid in South Africa, or might slavery explain police violence or the disproportionate incarceration of Black people in Brazil in ways that share mechanisms with the US (Arruda de Amaral 2024, 5)? If so, that would call for a broader theory of race, crime, and justice rather than merely viewing the US as singularly punitive.

Ellis Monk (2016) has argued that Brazil is a good comparison case to the US for studying ethnoracial inequality more generally, especially socioeconomic stratification by skin tone. A key reason to compare skin tone across these countries is because Black in one case is not necessarily Black in the other. Consequently, elevating skin tone as an analytic category provides a common basis for comparison across the two countries, and probably others. In other words, cross-national comparisons of race may be misleading when the nominal categories, such as Black, don’t mean the same thing across cases, calling for an analytic shift to color.¹²

In short, there may be more similarity than difference in mechanisms connecting racial inequality to punishment when viewed from an alternative lens of global comparison and by accounting for the differing crime challenges that countries face. Once again, this possibility underscores the need for integrating theories of crime and punishment. Yet the research base is limited, and many questions remain. For example, while homicide may satisfy measurement reliability concerns for some comparative purposes, it is quite rare relative to the overall crime problem, which is heterogeneous in nature and highly variable over time. Sexual violence, hate violence, nonlethal gun victimization, and violence against the state are relevant, as is social disorder and conflict. Gun violence in the US is also distinctive relative to other countries, uniquely contributing to its high homicide rate according to many scholars (Zimring and Hawkins 1999).

Rates of change rather than absolute levels are important as well, and of course other hypothesized causes of country differences in punishment such as racism and legacies of racial subjugation, localized politics, the political economy, and the decline of the welfare state (Lacey et al. 2018; Wakefield and Turney 2025) are ripe for deeper international comparisons, for explaining not just incarceration but all forms of penal control. Garland (2020, 321) argues that these forces are all intertwined and that “we have to consider America’s exceptional levels of punishment together with America’s exceptional levels of violence and disorder, while understanding both of these as outcomes of America’s distinctive political economy.” Few have done so, however, or tested alternative explanations directly when it comes to racial inequalities in crime and punishment. Even fewer studies have examined within-group racial inequalities in punishment by skin tone, especially comparatively.

Largely unanswered too is the question of whether the diffusion of punitive measures of control or surveillance to institutions not formally connected to the criminal legal system is greater or less in the US compared to other countries, again assuming we have answered the question of the right comparison. To assess whether carceral control has spread to adjacent institutions requires comparable measures over time and countries, often precisely involving institutions that typically do not provide the publicly available or reliable data. Although these measurement and theoretical challenges constitute a hard problem, advances are being made in creating comparable international databases for child welfare (Roehrkasse et al. 2023). In the US, the challenges are notably steep, the reason that Lee and colleagues (2025) call for a new wave of systematic data collection. In the meantime, comparisons of the sort they provide for the US form a template for applying to other countries. The availability of state registry data on entire populations, such as in Denmark and Finland (see, for example, Andersen 2017), may prove a boon for the comparative study of how much forms of punishment have expanded, shrunk, or remained stable in the recent past. The rise of the carceral state, both in the US and globally, is very much a live rather than settled research question.

STRUCTURAL RACIAL INEQUALITY AND LIFE COURSE DISADVANTAGE

According to the National Academy of Sciences, Engineering, and Medicine (2023, 22), structural racism “refers to a system in which historical and (or) contemporary public policies, institutional practices, cultural representations, and other norms work in different, often reinforcing ways to maintain or compound racial inequalities.” The advantage of this definition is that it invites a wider view than traditional work focusing on racial bias by individual agents of the criminal justice system. The disadvantage is that it is potentially too broad and vague (Wacquant 2023, 86), and that other labels, such as structural discrimination or even structural inequality, cover similar ground. As a result, the definition and especially measurement of structural racism is fraught with difficulty (Lee 2024).

A way forward is provided in this issue along with an emerging body of research that is disaggregating the component parts of racial stratification in concrete material conditions of life that influence crime and criminal legal processing, what I call structural racial inequality. Dovidio and Solomon (2025) take a multilevel systemic approach in assessing the considerable evidence of racial disparity and discrimination in the encounters of Black citizens with police, including but not limited to traffic stops and police shootings. In their evaluation, the implicit or even explicit racial bias of individual officers is secondary to systemic or supra-individual sources of bias that arise from structural sources like racial segregation. Patrol and policing practices (such as stop and frisk) and the perceptions of both citizens (who frequently call the police) and other criminal justice actors like probation and parole officers are shaped by the structure of how society is spatially produced and racially organized. It is well documented, for example, how bank redlining and restricted zoning housing policies by the government created persistent racial segregation by neighborhoods in America (Rothstein 2017). Decades of research have also shown how racial segregation concentrates both poverty and crime (Massey 1995), which in turn exposes residents to forms of aggressive policing, concentrated incarceration, and neighborhood stigma, creating a mass of correlated disadvantages in life for African Americans (Sampson 2024). Policing and the criminal legal system are embedded in this larger structure of racial stratification.

A related line of place-based research is probing how spatial boundaries shape social conflict and the racialized contact of citizens with the criminal justice system beyond the census-type definitions that researchers have traditionally relied on. Joscha Legewie and Merlin Schaeffer (2016), for example, find that 311 calls in New York City about social problems such as neighbors making noise and drinking in public are most likely to occur in the “fuzzy” transitional spaces located between racially homogeneous communities. Using stop and frisk data for the same city, Neil and Legewie (2024) also show that White neighborhoods along racial boundaries experienced higher rates of police stops independent of crime rates and compositional and socioeconomic characteristics. Further, they found that the police required less suspicion before conducting stops in White neighborhoods along racial boundaries, but only for Black and Hispanic suspects. Whether through neighborhood or spatial boundary effects, there is evidence that racialized social control leads to higher rates of police contacts with African Americans.

Bushway and colleagues (2025) focus on the criminal legal practices of the courts and legislatures that create disparate impacts for minority defendants without generating improvements in public safety, which is their definition of structural racism. What unifies their approach with Dovidio and Solomon (2025) and other articles in this issue is identifying a mechanism that does not require individual bias on the part of criminal justice decision-makers. In particular, Bushway and colleagues (2025) focus on racial disparities in court practices, including pretrial procedures, plea bargaining, sentencing, and community supervision. Many racial disparities can be explained by differences in case or defendant characteristics, like prior crime, but their theoretical framework and analysis highlight how structural factors like cash bail systems have created racial disparities by linking pretrial release to ability to pay, which has disadvantaged minorities, and how mandatory minimums have targeted drug crimes such as crack cocaine, which has also disadvantaged minorities. Neither of these policies has been shown to reduce crime in a meaningful way. Their key conclusion is that penal practices such as systems of cash bail, unchecked discretion in probation and parole revocation, mandatory minimums, and “three strikes” legislation likely contribute more to racial disparities than direct discrimination. As Bushway and colleagues (2025) note, conditional on being processed in the criminal legal system, racial disparities in case outcomes stem mainly from the punitive nature of numerous practices, regulations, and laws.¹³ This is an important claim with clear and immediate policy relevance—reforming these structures could significantly reduce disparities even without major changes in other aspects of society and without increasing crime. As they reveal, in some cases policy changes have already significantly reduced racial disparities in sentencing.

One can also apply a structural inequality perspective to the life course, both before and after criminal legal contacts. Nahra and colleagues (2025) explore life after prison, a venerable research tradition in the field of criminology but one that has been renewed by the reach of mass incarceration. Their move is to conceptualize how to repair the fallout from the rupturing institution of prison that cuts inmates off from the social and material ties that bind. The process of reintegrating back into society is especially troubled for individuals cycling through a series of compounded disadvantages and violence (Miller 2021; Western 2018).

The fraught history of racial inequality in America and its intergenerational transmission mean that African Americans in particular face the greatest burden of reoccurring disadvantageous circumstances (Sharkey 2008). As noted earlier, disadvantages like poverty are concentrated spatially and cluster with many other neighborhood adversities like joblessness, violence, and disinvestment (Sampson 2024; Sharkey 2013), and research on high-imprisonment neighborhoods and high-imprisonment families shows that networks of criminal activity and criminal justice contact are often deeply enmeshed (Western 2015). Further investigation is needed into whether community institutions like churches, schools, and nonprofit organizations mitigate or amplify these more structurally rooted inequalities. Racial inequalities in exposure to correlated disadvantages also cut across multiple individual, family, and neighborhood contexts, creating an assemblage of adversities that cumulate over the life course (Desmond and Western 2018, 308–09).

Nahra and colleagues (2025) provide a way forward for policy and research, not just on reentry back into society from prison, but on the idea of desistance more generally. Traditional research on desisting from crime, one that criminal justice policy has long followed, focuses on official indicators of recidivism, like arrest. But as we have seen, these are problematic indicators due to racial inequalities in policing. The authors provide a road map for efforts to redefine desistance, as does Monica Bell (2021) more generally when it comes to envisioning and defining safety. Equally important, the process of social integration, which rests on supportive social relationships, secure housing, employment, and physical and mental health, needs to be prioritized in measurement analytics and in policy support. As Nahra and colleagues (2025) argue, both criminalization and punishment can weaken one’s ties to family and community. The increasing complexity that defines America’s families has additional consequences for promoting desistance in the reentry process, both concurrently and intergenerationally (Wildeman and Sampson 2024). Noncrime policies in health, housing, and employment are therefore crucial for supporting life course processes of reintegration in the transition from prison to community.

Taking a structural, life course perspective opens up other lines of inquiry as well, including assessing how cumulative inequalities early in life lead to lifelong disparities in criminal justice contacts and how these vary by race. Sampson and Neil (2024, 183) hypothesize that the differences between racial groups in cumulative exposure to multiple social advantages and disadvantages—at the individual, family, and neighborhood levels—explain racial group inequality in the experience of arrest over the life course. This perspective draws in part on community-level theories of racial differences in crime rates, while simultaneously emphasizing the historical and structural reasons why minority neighborhoods, especially predominantly Black neighborhoods, face elevated rates of concentrated disadvantage (Massey and Denton 1993; Sharkey 2013; Crutchfield 2022, 120; Bellair and McNulty 2005), a perspective that is consistent with theories of structural or systemic racism (Bell 2021).¹⁴

Using multi-cohort longitudinal data on nearly one thousand individuals from Chicago, Sampson and Neil (2024, 177) find that large Black-White disparities in arrests from people ages ten up to forty can be attributed to racial inequalities in exposure to cumulative childhood advantages and disadvantages rather than race-specific effects. Hispanic-White gaps in the incidence of arrest, though smaller, follow a similar pattern, and the same patterns hold across different offense types and across birth cohorts who came of age at different times during the period 1995 to 2021. These findings support the argument that racial inequalities in early-life structural factors, which themselves are historically shaped, trigger processes of cumulative advantages and disadvantages that produce racial disparities in arrests over the life course and that persist across different points in contemporary history.¹⁵

To be sure, the cumulative pathways shaping racial disparities in arrests do not necessarily run through criminal behavior. Police officers and parole agents, for example, are responsive not only to criminal behavior but also to perceived race and other cues such as neighborhood poverty and family social class (Kirk 2008; Smith 1986). Persistent exposure to intensive policing practices that target individuals based on perceived race, though ultimately stemming from the history of race and racism, can directly influence arrest. Yet racial inequalities in early-life social conditions can produce racial disparities in arrests over the life course independent of direct racial biases among individual officers, in part through producing differences in criminal behavior and in part through producing differences in exposure to institutionalized policing practices that disproportionately harm Black communities.

It follows that racial inequality shapes disparities in both criminal behavior and arrests through a stratified life-course process. As Sampson and Neil (2024, 200–201) argue, historical inequalities and intergenerational linkages produce divergent early-life social contexts by race, which through general processes of cumulative advantage and disadvantage go on to manifest as long-term inequalities. This conceptualization distinguishes race in the moment of arrest from how race structures the life course leading up to that moment. Processes of cumulative disadvantage begin at birth and involve exposure to an array of disadvantages that are not only spatially concentrated but also cross-cutting across individuals, families, and other settings that have long been subjected to structural racial inequality.¹⁶ The implication, as Robert Crutchfield (2022, 124) argues, is that the comparative life chances of Blacks and Whites are shaped not only by being arrested and imprisoned, but also by the “criminogenic conditions of their lives.”

In combination with the arguments of Bushway and colleagues (2025), Dovidio and Solomon (2025), and Nahra and colleagues (2025), it follows that even if racial bias in policing and criminal justice processing were somehow magically eliminated tomorrow, that would not necessarily portend an elimination of racial disparities in criminal legal contact or its collateral consequences if there remain structured inequalities by race in law, social contexts, and life chances that lead to criminal behaviors subject to punishment. This is an underappreciated point that has important implications for understanding and potentially eliminating racial disparities in criminal justice.

CONCLUSION

Absolute and relative disparities in incarceration by race have declined significantly across the United States in the past fifteen years, a trend that has surprised many observers. In Massachusetts—my home state—the oldest men’s prison recently closed (Lam 2024), and incarceration rates have reached historic lows. Does this signal the end of mass incarceration? Perhaps, but in states like Texas and Louisiana, incarceration rates are still high and have again started climbing.¹⁷ In addition, racial, ethnic, and immigration-related disparities in incarceration persist nationwide and may be worsening.

These varying trends expose the reality of ongoing shifts, sometimes over short periods. Indeed, we have repeatedly witnessed the mark of time imposed by large-scale social changes involving crime and criminal legal processing, which have challenged researchers who attempted to predict them (Sampson, forthcoming). Neither extreme pessimism nor extreme optimism accurately represents recent realities. Although we have grounds for optimism that racial disparities in incarceration and the broader criminal justice system may be receding, there is no guarantee these improvements will last. As this issue demonstrates, racial inequality in the US criminal legal system remains both persistent and dynamic.

A new research agenda emerges from this framework. Foremost, we must develop clear conceptual and empirical criteria for examining changes in criminal justice processes and linking them to broader societal trends, integrate our understanding of the social causes of crime into analyses of formal social control, and explore institutional and structural factors that produce racial inequality, in addition to recognizing individual-level bias in the criminal justice system. Although the next generation of research must address these needs, the evidence supports a dual policy approach in the present that simultaneously reduces structural racial inequality in laws and in life conditions—whether individual, familial, neighborhood-based, organizational, or societal. One approach offers more immediate effects, while the other takes a longer view. We should embrace both.