Imagine it’s a warm and sunny day. While walking back to your office from your favorite lunch spot, you see a street vendor selling strawberry ice cream. You feel in your pocket for change to see if you have the $1.50 to buy a cone; to your surprise, you discover 6 quarters. With a smile on your face and the sun on your shoulders, you purchase your cone and stroll back to your office, taking a detour through the park as you enjoy your ice cream.
The next day you receive the following email from the ice cream vendor (you were in such a good mood you agreed to provide the vendor with your email address to receive future coupons):
Dear Strawberry Ice Cream Fanatic,
We would love to offer you a free and delicious strawberry sundae next time you pass by, if you first help us by answering the following questions:
• How likely are you to buy strawberry ice cream from us in the future?
• How likely are you to buy strawberry ice cream from anyone in the future?
• What factors went into your decision to buy strawberry ice cream from us yesterday?
• What factors will go into your decision to buy strawberry ice cream from us in the future?
Strawberry Dreams Forever,
Your Favorite Strawberry Ice Cream Vendor
Your first thought is, “Wow, these people are serious about the strawberry ice cream business. I haven’t the faintest idea as to what all went or goes into my strawberry ice cream-purchasing decisions, I just felt like strawberry ice cream yesterday. How do I even answer this?” But you do really want that free sundae, so you quickly cobble together some answers and send the email on its way.
So, what does this strawberry ice cream anecdote have to do with probation? The ice cream vendor’s email is a bit like a tool commonly used in probation—the risk and needs assessment (RNA). Like the email from the ice cream vendor, RNAs seek to gather data with the goal of predicting future behavior—only in probation we’re trying to predict whether and under what circumstances a person will reoffend. It goes without saying that the stakes are much higher than trying to predict whether you’ll buy strawberry ice cream again.
But the anecdote serves another purpose: The ice cream vendor’s email asked the right questions, but it didn’t necessarily produce accurate answers, because you just quickly provided some information (or even made something up!) to satisfy the request and earn your free sundae. The vendor, however, doesn’t know that, and may blindly move ahead with business decisions based on your answers. Similarly, an RNA can ask the “right” questions and yet still fail to accurately predict criminal recidivism or to accurately assess criminogenic needs.
That is not to say that RNAs are inherently inaccurate, simply that using them can lure probation officers into feeling as though they have a definitive answer about the probationer’s risk of re-offending when in fact the situation is much more complex. RNAs are useful tools, but they do require a healthy dose of caution.
Know Your RNA & Don’t Rush the Process
RNAs vary in complexity. All probation officers using RNAs should receive training on the specific RNA being used, including the types of offenders for which the RNA was developed, how to administer the RNA and what decisions are appropriate to make based on data produced by the RNA. If you’re being asked to use an RNA, make sure you have the appropriate training first.
Taking your time with the RNA is equally important. Probation officers are often expected to complete the RNA quickly as part of the intake process, or to work with the results of an RNA administered by another officer. Sometimes the RNA is mandated and must be completed by a certain deadline, which can add to the pressure.
It’s important not to rush the process or complete the RNA without necessary information. An inaccurate assessment not only has potential negative consequences for the probationer and the community; there’s risk to probation officers and agencies, too. It is not unheard of for litigation to arise regarding negligent supervision based in whole or in part on an improperly administered intake and risk assessment.
RNAs can lure probation officers into feeling as though they have a definitive answer about the probationer’s risk of re-offending.
Know the Limitations
Like any tool, you’ll be able to use RNAs more effectively when you understand their limitations. Those include:
- We don’t always know how they work. Many RNAs are developed by for-profit companies that do not reveal their algorithms. In a 2016 case, a defendant challenged the use of the COMPAS score in his sentencing, arguing its accuracy can’t be validated because it was developed by a private company that considers the information proprietary. In its ruling, the Wisconsin Supreme Court acknowledged this limitation, ruling that COMPAS scores could be used in sentencing, but the judge had to receive written warnings about the value of such scores and couldn’t rely on them exclusively.
- They use algorithms that may involve racial bias. RNAs purport to take bias and opinion out of the sentencing or probation process and replace it with an objective, neutral assessment. Unfortunately, algorithms do not always produce neutral results and numerous studies have shown artificial intelligence to have bias and prejudice. ProPublica examined the RNA scores for more than 10,000 people arrested in Broward County, Fla., then compared them with those individuals’ arrest records over the next two years. The researchers concluded black defendants were twice as likely to be incorrectly labeled as higher risk than white defendants.
- They can score some offenders high-risk or low-risk based on factors that don’t have a lot of relevance to their crime. As a prosecutor, I saw many RNA scores that were based on 10 questions, one of which is age. A sex offender over the age of 50 would often get rated low-risk, because the RNA was designed to assign much lower risk to older offenders. But probation officers and prosecutors know sex offenders are notoriously risky. They may stay out of trouble for a while, but they often revert. It can work the other way, too. In that same ProPublica study, only 20 percent of the people the RNA predicted would commit violent crimes went on to do so. This is probably why it’s common practice for probation officers to override sex offender RNA scores.
Reassess as Conditions Change
An RNA provides a static data point at a particular time in a probationer’s life. But probationers are just like the rest of us—their lives change over time. A probationer who loses his job may be at much higher risk for drug abuse than when he was originally evaluated. The RNA may have indicated no need for drug abuse counseling when first administered, but now the need is very real to keep the probationer on track. Probation officers must be on high alert for these life changes and reassess the offender’s risk and needs accordingly, adjusting the level of supervision or conditions of probation as needed and seeking court assistance when necessary.
Incorporate Other Information
RNAs should not be the only basis on which probation officers assign the probationer’s level of supervision. Additional information to consider includes:
- Court-ordered directives related to supervision
- Mental and physical health evaluations
- Substance abuse evaluations
- Information collected at intake
- Nature and severity of the offense requiring supervision
- Past criminal history and past performance on probation/parole supervision
Use Your Discretion
With overwhelming caseloads, probation officers need risk assessment tools like RNAs. They provide an important data point in the effort to identify the risk for criminal recidivism and assign an appropriate level of supervision. But the factors that influence whether someone will re-offend are inherently complex—and miscalculating carries enormous potential risks for the probationer, the community and the probation officer. When using risk and needs assessments in probation, it’s important to understand the limitations of the tool and use your discretion. After all, your decision is far more important than the ice cream vendor’s prediction of what might lead you to purchase strawberry ice cream on a sunny day.