Judges are trying to better integrate empirical evidence into their courtrooms, but a newer policy — evidence-based sentencing — is coming under increasing scrutiny from civil rights advocates.
Through evidence-based sentencing, various socioeconomic and demographic factors are used to determine the length of a defendant's sentence. These can range from more controllable factors like a defendant's criminal record to things like the rate of violence in a defendant's neighborhood or even a family member's criminal record. These measures are entered into a computer program, which spits out various models that judges use to determine a sentence for a defendant.
The problem, as US Attorney General Eric Holder recently argued, is the measures could disproportionately hurt the poor and minorities. "I'm really concerned that this could lead us back to a place we don't want to go," Holder said in an interview with TIME.
Sonja Starr, a law professor at the University of Michigan, and co-director of the Empirical Legal Studies Center, said that the inclusion of demographic and socioeconomic factors in these models is "unfair." In an interview, Starr, who previously criticized evidence-based sentencing in the Stanford Law Review and the New York Times, broke down the problems with the policy, and how a system created to reduce sentences might do the opposite — at least for certain segments of the population.
German Lopez: What's your general view on evidence-based sentencing?
Sonja Starr: Evidence-based sentencing is being used through these risk-prediction instruments that try to predict whether the defendant will commit a future crime. When these instruments include, as so far they always do, factors that are inherent to the individual, like demographic or socioeconomic factors, I believe they're unfair and we should reject them as policy. I also think several of the factors are unconstitutional for sentencing judges to consider.
GL: How specific do these factors get?
SS: None of the instruments in effect currently consider race. They do consider a lot of other nitty-gritty factors. For instance, many of them consider gender, age, marital status, unemployment or employment status.
Some of them are fairly simple. They only have eight to 15 factors.
But the most popular ones these days are more sophisticated and include a pretty detailed analysis of a lot of aspects of the defendant's life. They'll include, for instance, a financial profile that looks at his or her ability to pay bills, income, current employment, past employment, and so forth. They'll also include a family profile, which note unstable childhoods, whether someone was in foster care, and particularly if someone's family was involved in the criminal justice system. Some of them even penalize defendants when their family members are crime victims.
So there's a lot of analysis of factors that are quite unrelated to criminal conduct, although they do also include the defendant's criminal history.
GL: One of the things I've read a lot about is the school-to-prison pipeline. It seems like some of these factors might worsen that problem, and kids with convicted fathers will be even more likely to spend time in prison.
SS: These instruments themselves take that fact into account, and they treat it as a self-fulfilling prophecy as a justification for additional punishment.
It is true in our society that poverty is generally predictive of violent crime risk. Not just that poor people are more likely to go to prison, but the social causes of crime are often rooted in poverty. So these instruments take that fact and they use it to say that once someone commits a crime, we should more scared of them committing another crime in the future if they're poor — because statistics show that, on average, they'll have a harder time escaping from that path of crime and they may pose more of a risk.
In my view, that's not fair. It generalizes all poor people, and it treats individuals differently than others who have done the same thing based on these group-based statistical averages. That's something that even the Supreme Court has warned against.
GL: A lot of these factors also correlate with being a minority.
SS: Even though these instruments don't include race, they do include a lot of variables that are closely correlated with race — including basically every socioeconomic factor.
Some of the instruments include characteristics of the defendant's neighborhood, such as the neighborhood's crime rate. Because we have a high level of residential de facto segregation in the United States, factors correlated with the neighborhood will tend to have racially disparate impacts as well.
GL: To my understanding, evidence-based sentencing was conceived with the hope of reducing, not increasing, sentences. Do you think that's how the policy will always play out?
SS: Well, different states are looking to use evidence-based sentencing in different ways. Some states are incorporating the instruments as part of diversion from incarceration programs, so it's more explicitly for the purpose of reducing sentences. Some of them are just incorporated routinely into every pre-sentencing investigation report.
When you incorporate them into every report, it's quite certain that, even if you're hoping it will mostly be used to reduce incarceration, it's not only going to have that reduction effect. It's probably going to reduce incarceration for people who are low-risk and increase it for people who are high-risk. A judge isn't, all other things being equal, going to want to let someone go who has a high-risk assessment.
In the case of diversion programs, there might be reduced incarceration. But, while I think we do need to reduce incarceration, we should do so on the basis of assessments that don't use these problematic variables that make it so only the relatively privileged qualify for diversion programs.
GL: Do you think state judges might be under political pressure to look tough on crime through these programs, since they're generally elected?
SS: I think that's right. It's partly that judges genuinely worry about future crime victims, and it's partly that they may face political consequences if they let someone go who had a high-risk assessment and then they commit another crime.
When prosecutors see the reports, they also have a lot of influence over what somebody's sentence is going to be. So I think prosecutors will also push hard at the sentencing stage against defendants who have high-risk assessments.
GL: Is there a case that this whole system punishes people before they do something wrong just because they might do it?
SS: I do get the creepy-factor criticism.
I want to say, though, that it is traditional in sentencing for judges to consider the defendant's crime risk to be one of several important factors to take into account. Primarily, sentencing is based on what you deserve for the crime. But it's always been the case that protecting the public is one of the reasons we have a criminal justice system.
So I don't want to go too far. I don't want to say that thinking about the risk the defendant poses in the future is never relevant to sentencing. Instead, I think judges should base their risk assessment on the defendant's conduct. That is, if the defendant has a string of crimes already, you have good reason to think the defendant might commit more crimes in the future. But they shouldn't use these other factors that I view as inappropriate.
GL: Even in the states where evidence-based sentencing is used to reduce sentences, do you still see it as a problem that could affect certain segments of the population?
SS: Yes. In some of these states, if you have many of the risk factors, you essentially cannot qualify for diversionary programs. I don't think that's good.
One of the things that allows mass incarceration to continue in this country is the political insulation factor. The impact of mass incarceration is not equally distributed. The relatively privileged are more insulated from mass incarcerations and don't see its costs.
I think it's very important to think about strategies to reduce mass incarceration. But when we choose a strategy for getting there that further insulates the relatively privileged and provides nothing for those that have the wrong demographics or the wrong socioeconomic status, it just further worsens the problem.
Now, I don't think that the people who have advocated these instruments are evil or against the poor. They're well-intentioned people who are looking for ways to reduce incarceration, and that is an important goal. But I don't think this is the path we should take to get there.
GL: What are some ways, in your view, to reduce that insulation?
SS: I think recently we've had some efforts to try bring mass incarceration under control, because the sheer cost of the criminal justice system is now getting people's attention.
In the case of evidence-based sentencing, I think the transparency issue is one reason this hasn't gotten a lot of attention. These instruments are described in a way that makes them sound dry and difficult to oppose. They're evidence-based, they're scientific, they're based on regression studies.
People don't like to say they're against smarter sentencing or more informed sentencing. They might not really understand that when you give a risk prediction that's based on group-based averages you are essentially saying that the state should penalize for being black, male, young, unmarried, or having parents that went to prison.
GL: Would you consider more transparency a good reform even if evidence-based sentencing isn't fully repealed?
SS: Definitely. My hope is that if they did speak clearly about the metrics, they would have a harder time defending them. In truth, I doubt that very many policymakers or judges would be willing to go to the public and say, "I think that poor people should just go to jail for longer because they're poor and the poor are dangerous." That is a rationale that we have incorporated into our sentencing system in this very hidden, technical way by burying it in these risk assessments. If it had to be brought into the open, though, I don't think very many people would be willing to stand by them.
GL: Do you think these predictions are even accurate in the first place?
SS: They're not very accurate when it comes to individuals. They are statistical generalizations that are true when applied to large groups. That is at best if the study is done well.
Group-based averages are really only good at predicting what happens in future groups as a whole. They will be wrong in lots and lots of cases when it comes to particular individuals. There's a ton of variation across individuals.
GL: What do you think would be some good ways for judges to assess risk among defendants?
SS: Judges should think about doing more data-driven predictions, given that judges always engage in some kind of prediction of future behavior.
If it's done, it should be based on the defendant's conduct. What was the crime in this case? What kind of crime did the defendant commit in the past? And maybe other factors that are changeable characteristics, like whether the defendant expressed remorse for the crime.
But basing it on these fixed characteristics that are outside of the defendant's control is what we should avoid.
This interview has been edited for length and clarity.
To learn more about the criminal justice system's racial disparities, watch the two-minute video below: