Skip to main content

Data-Driven Decarceration

A close analysis of prison data can help us think strategically about the tradeoffs of different approaches to ending mass incarceration.


Mass incarceration is so deeply entrenched, it takes tremendous energy just to get the decarceral ball rolling. But deciding how to channel this momentum once it builds can also be a real challenge. How, exactly, should we shrink, unwind, or close prisons?

No one expects this work to happen all at once. Still, it can be helpful to identify and assess decarceral pathways at the outset. The options are numerous, and even those that would achieve the same numeric target—whether it’s reducing the prison population by 10, 50, or 100%—can have vastly different social consequences. Some approaches, for example, might be structured to divert as many people as possible from ending up in prison in the first place. Others might try to minimize the risk that decarceration could increase crime on the outside. Still others might seek to reduce the prison population as quickly as possible. And still more might focus on limiting racial disparities among those still incarcerated.

No one path to decarceration can achieve all these goals at once. That’s in part because decarceration—by which I mean the by-the-numbers work of reducing the prison population—would likely occur against a backdrop of familiar constraints, including the composition of today’s prison population and the political reality of our current social order. Even when decarceral reforms are in reach, in other words, we may need to consider the tradeoffs of competing decarceral pathways and make choices that reflect deep value judgments about our priorities.

To make these choices more concrete, in a recent article I used prison data from 39 states to forecast the effects of hundreds of decarceral strategies. My goal was to answer a few key questions: What would happen if we cut prison admissions—by decreasing the number of convictions or carceral sentences—by 25, 50, 75, or 100%? What would happen if we shortened the time prisoners serve—by cutting sentences or granting clemency—either retroactively to current prisoners or prospectively to future ones? And what if we combined these strategies or applied them to some offenses but not others?

My main findings don’t hinge on any particular numeric target—whether we aim to reduce the prison population by 10 versus 50%, for example, or to zero it out altogether. Rather, my goal is to identify strategies that reach the same target and explore the competing tradeoffs they represent. To keep things simple, I’ll assume the goal is cutting the prison population in half, but the same basic insights apply to other goals too—even shuttering every prison in the country. Taken together, my forecasts measure a few key tradeoffs, which highlight hard choices in scaling down the carceral state.

Fast or Slow

We cannot talk about decarceration without also talking about timelines. With the right political will, some strategies could happen almost overnight. Universal clemency, for example, could bring the prison population to zero immediately. By contrast, reducing admissions or the length of time served by future prisoners could take decades to realize: Even if we stop convicting people entirely or slash sentences for all future cases, the prison population will remain large for years because many people are already serving long sentences. These polar examples highlight how the structure of a decarceration strategy impacts its timeline.

The same lessons apply to less sweeping reforms. For example, if we cut sentences by 75% for future prisoners convicted of nonviolent or less-serious violent offenses (such as robbery and assault), we could reduce the prison population to half its current level—but it would take more than a decade. By contrast, applying similar reductions to both future and current prisoners would achieve the same goal immediately.

More from our decarceral brainstorm

Every week, Inquest aims to bring you insights from people thinking through and working for a world without mass incarceration.


Sign up for our newsletter for the latest.


Given the fickle dynamics of U.S. politics, quicker decarceration has real advantages. Take Arkansas for example. In 2013 the state’s prison population was declining due to legislation enacted just a few years earlier. But after someone on parole committed a heinous murder, key reforms were backtracked under public pressure, causing the state’s prisons to grow again. As Arkansas illustrates, slower decarceration strategies allow more time for political pushback, and thus may leave more people by the wayside.

That said, quicker decarceration also has its own political costs. Perhaps most important, some government officials may be unwilling to support a strategy that they perceive provides insufficient insulation against political backlash, as in Arkansas. Slower decarceration can push the risk of high-profile crimes to the future, sometimes after the relevant decisionmakers will have already retired or left office.

This timing tradeoff is most intense for violent offenses because they carry the longest sentences. If applied to violent offenses, the slowest decarceral approach—shortening time served for future prisoners alone—would take over 30 years to fully realize. And only a third of that effect would occur in the first five. The key point is that slower strategies can delay the bulk of decarceration’s effect, which could help bring policymakers on board in the first place. But it also provides ample time for opposing political forces to push back before the project is completed.

Two Types of Harm

In seeking to reduce one form of harm—civilian-on-civilian crime—prisons inflict other serious harms on the same communities they purport to protect. Incarceration imposes enormous costs on prisoners and their families, particularly among Black people and other people of color. In the words of legal scholar Allegra M. McLeod, prison “degrades” individuals’ “basic humanity and sense of self-worth” through invasive physical controls. It disrupts education and careers; separates families; exposes prisoners to disease, violence, and trauma; disenfranchises; and, for some people, it may even be criminogenic.

Fortunately, there are ways to reduce crime without the same social harms. As abolitionists have argued for decades, the best approach is likely large-scale investment in communities and infrastructure. And, as I’ve noted elsewhere, there’s promising empirical evidence that violence is reduced by investments in early childhood education, health care (including mental health), substance abuse treatment, cognitive behavioral training, violence-prevention programs, lead mitigation, emergency cash payments, greening of vacant lots, and even street lighting.

Critically, however, these community interventions require time both to implement and take effect. Their impacts on crime will therefore lag. Based on the available empirical evidence, many scholars believe that some carceral approaches that are already in place—like policing and incarceration—decrease civilian-on-civilian crime, at least under certain circumstances. In the short term, then, a shift from carceral to non-carceral crime-reduction policy risks increasing crime.

Efforts to reduce the prison population thus face a near-term tradeoff between two types of harms—those caused by crime and those caused by prison. Strategies that shrink prison while seeking to minimize crime tend to focus on people convicted of low-level offenses, presumably because they are thought to commit less crime upon release. They also emphasize shortening time served, rather than decreasing admissions, because (admittedly contested) theories of punishment like deterrence and incapacitation predict that any crime-reducing effects of incarceration diminish with each additional month of a prison stint.

But shortening time served does nothing to divert people entirely from the harmful conditions of prison itself. To do that, we must instead reduce admissions. And, given some numeric target for reducing the prison population, we can minimize admissions by focusing on offenses with the shortest sentences.

This tension between the harms of crime and the harms of prison contact is exacerbated by a critical fact: We cannot shrink the prison population, even by half, by decarcerating nonviolent offenses alone. Only 14 and 28% of people in state prisons are serving time for drug and other nonviolent crimes, respectively. Thus, to make large-scale inroads on mass incarceration, we need to include violent offenses, too—a statistical reality that highlights the danger of focusing decarceral efforts exclusively on nonviolent offenses and thus reinforcing the perception that violent offenses are too dangerous for decarceration.

Once again, the tension between the harms of crime and of prison become more concrete in the forecasts. Let’s assume we continue to use only our current carceral policies for crime reduction. If we seek to cut the prison population in half while minimizing the short-term risk of crime, one potential strategy would shorten time served by 75% for both nonviolent and less serious violent offenses. Alternatively, if we care most about diminishing the social harms of incarceration, we could instead eliminate all admissions for nonviolent offenses and reduce admissions for less serious violent offenses by a quarter.

Both options would cut the prison population in half. But, by shortening time served alone, the first continues sending the same number of people to prison each year. The second, in contrast, lowers admissions by over three-quarters, diverting over 300,000 people from exposure to the harmful conditions of incarceration annually.

Race on the Inside

Racial disparity is a defining feature of mass incarceration. In 2020 Black and Hispanic people in the United States were incarcerated at a rate of 938 and 446 per 100,000, respectively. For white people that number was just 183. These disparities are even starker when broken out for gender and markers of social class. For example, in 2007, 26% of 18- to 30-year-old Black men without a high school degree were incarcerated.

Ending racial disparity is a central aim of many decarceral efforts. My forecasts, though, highlight a dilemma: Virtually all plausible pathways would exacerbate or leave unchanged current levels of disparity.

This finding stems from the fact that, relative to white prisoners, Black prisoners are admitted at heightened and roughly constant rates per capita across offense types—with the exception of violent offenses, for which they are admitted at even higher rates. It also stems from Black prisoners serving, on average, a similar length of time as other prisoners for nonviolent offenses but more time for violent ones. The upshot is that decarcerating nonviolent offenses alone would likely increase racial disparities among those left behind—by up to 13%, according to my forecasts.

Out of all the decarceration strategies I tested, the only ones that reduce racial disparity are those that decarcerate violent offenses far more than nonviolent ones—an unlikely policy reform. And, even then, they never reduce Black overrepresentation dramatically.

To be clear, this is not an argument against decarceration, nor is it an argument that decarceration is inconsistent with racial justice. After all, Black prisoners would disproportionately benefit from nearly all decarceration strategies precisely because they are so overrepresented in prison. But the fact remains that virtually all plausible efforts to shrink the prison population would leave us with similar or even starker racial disparities among those who remain.

Reducing Violence

My findings about racial disparity suggest that decarcerating violent offenses provides an opportunity to both contract prison and limit racial disparity. So how could we do it?

As noted, one approach is to make large-scale infrastructural investments, especially in Black and brown communities. We can also shorten overly long sentences and decrease admissions for at least some violent offenses, particularly less serious ones—like robbery and assault—for which the United States imposes far more prison time per offense than other comparable countries.

Developing the political will here is a real challenge, but some jurisdictions are already doing it. My article outlines a variety of strategies, including finding ways to communicate to policymakers and the public that people convicted of violent offenses are rearrested for a violent offense only slightly more often than people convicted of other crimes.

It might seem obvious, but decarceration isn’t one thing. There are many ways to reduce prison populations, each with important tradeoffs that would help and hurt different people at different times and in different ways. Grounding these distinct paths in data makes them more concrete. It also helps illustrate how decarceration and quantitative empirical research—sometimes thought to be in tension—can benefit and reinforce each other.

Image: Adam Birkett/Unsplash