Risk Assessment Tools: How Science Is Making It Safe to Release

In a quote oft attributed to Mark Twain, “It’s not what you don’t know that gets you in trouble, it’s what you think you know that just isn’t so.” And for many, there’s a lot we think we know. Most people could probably recite a Miranda warning. How? Because we’ve seen it on television. As it relates to bail, TV has shown us bounty hunters tracking down absconders. Who hasn’t smiled while watching a TV perp’s crestfallen look when the judge triumphantly declares “Bail denied!”?Bail is supposed to be set with the public’s safety in mind, a notion repeated by the proponents of the current money bail system such as the California Bail Agents Association.

We must be careful not to confuse the wonders of Hollywood production with real life. Indeed, after a lifetime of watching NYPD Blue, CSI, and a litany of other shows known by initials, one may think the judge has some whiz kids of their own on staff. Why not, if law enforcement can pull a defendant’s entire “file” being in seconds, wouldn’t the judge be able to do so as well? The truth is, when setting bail the judge may have little more than his or her “gut,” and the charges in front of him, which for bail in California must be considered as true.

Moreover, with pre-set county-wide bail schedules , someone can post money bail and be released before even seeing a judge. High-risk individuals who may pose a significant danger to public safety can be quickly released, while low-risk non-violent ones remain in jail for longer periods. A better solution than money bail and the non-refundable 10% deposit is science. Using science, we can replace the money bail system, safely release more people (who still haven’t been convicted), keep our communities safe, and save money. This post touches on one such empirically-based method known as Risk Assessment (RA) tools.

Risk Assessment Tools- Replacing the Proverbial Gut.

RA tools are objective research-based instruments, using risk factors that can provide a dependable prediction of whether a person released from jail will (1) fail to appear, or (2)  be a danger to the community. They make an objective assessment looking at the person, not just the crime, and will provide a classification based on risk. Subjects are rated “low risk,” “moderate risk,” or “high risk.” While RA tools look like little more than a survey, these tools have been accurately predicting risk for four decades with systems from Kentucky to Virginia.

Where RA tools have been implemented, they have been wildly successful. More people are released, which equates to less money spent detaining people, as well as solving the ethical quandary of locking up people not yet convicted of crimes. Moreover, rates of non-appearance and re-offending have not gone up in meaningful ways. There are several models in place across the U.S., but the first (and most famous) is the Virginia RA tool.

Virginia’s RA tool was created in 1998 and for over 10 years has been implemented state-wide. Since implementation, the Virginia RA tool has collected data demonstrating that 25% more defendants could be released without increased risk of failure to appear or increasing danger to the community. Moreover, this tool has been validated.  (Validation means an audit that looks to whether the factors measured are predictive.) Over time, some factors have been added and others removed. In the validation study, researchers found success rates reached as high as 87% for low-risk groups, and 64% for the highest-risk groups. These high success rates for lower-risk individuals are not an anomaly: a similar system in Colorado (pg 12 of pdf) duplicated the same 87% figure. By using these statistical tools more individuals can be released, while maintaining high likelihoods of success (defined as appearing in court and not violating any ordered conditions of release.)

While no two RA tools are the same, they all ask similar questions. The following factors have been validated by multiple studies,

  • Prior failure to appear
  • Prior convictions
  • Employment history/status
  • History of drug abuse
  • Additional pending cases
  • Severity of the current offense.

Other factors that are supported by research but which have not yet been widely validated are:

  • Status of active community supervision at time of arrest (probation, parole, etc.)
  • History of violence
  • Stability of housing
  • Community ties
  • Caregiving responsibilities to family

Putting It All Together

Once the RA has assessed a defendant’s risk level, that information can be used to ascertain what set of conditions (if any) can mitigate the risk that the defendant will not show up to court or be a danger to the community. A report may be submitted to the judge with recommendations for release with certain conditions, based on individualized risk and not just the instant offense. Often, this investigation and any subsequent supervision are performed by an agency called Pretrial Services (discussed in my next blog post). Making this type of personalized recommendation is something a bail schedule simply cannot do.

The list of restrictive/preventative conditions available to a judge is quite expansive. It may sometimes be set in policy, or in statutes themselves. They can vary from simple weekly phone calls, required visits, drug counseling and testing, to curfews and house arrest. However, it is unwise to simply “throw the book” at every defendant. These conditions are not appropriate for every individual, and more intensive ones are expensive and disruptive if used improperly.

Why “Throwing the Book” is a Recipe for Failure

Studies have repeatedly shown that while moderate and high-risk individuals benefit from more restrictive conditions, with lower risk individuals the opposite is true; excessively burdensome conditions make failure more likely. Compare the two following examples, where both people are arrested for possession of drugs.

Criminal 1 is a student arrested with a gram of cocaine. He does not use drugs but  has“held” some questionable packages for a friend for extra cash. He has a car, a roof over his head, food in his fridge, no history of substance or mental health abuse, and a job (which when coupled with school, means he actually has very little free time to get into trouble). Placing him in a halfway house would rip him away from his home, family, and friends—a stable environment. It would impose drug counseling and treatment for an issue he does not have, and add more release conditions making him more likely to fail and violate. And, at a cost of $73 a day (only $7 cheaper than incarceration in the Federal Bureau of Prisons), this condition burdens both the tax payer and the individual who doesn’t need it.

Criminal 2 is arrested with a bag of methamphetamine and a glass pipe. He is homeless, unemployed, and has a history of substance abuse and mental health issues going back decades. As a result of his circumstances, he also has a wealth of free time on his hands. Placing him in a halfway house would have the opposite effect as Criminal 1. Criminal 2 would receive order and structure. He can receive treatment for his substance abuse and mental health issues. He will have a roof over his head and food in his belly. Slowly, he can begin to get his life back together, and his situation can be improved.

Despite what the proponents say, there is great doubt that the current money bail system is working as promised. Moreover, RA tools are viable alternatives that are empirically based. They have shown for decades that treating defendants as individuals, we can look at their unique circumstances, figure out what level of risk they pose, craft conditions to fit the circumstance and all while maintaining relatively high success rates.  While not perfect, RA’s are a valid alternative to the current system which permits the release of defendants based on the ability to pay, or a judge’s gut, but not any specific articulable facts.