WIXLIB

Issue BriefPretrial Risk Assessment 101:Science Provides Guidance on Managing DefendantsEvery day, criminal justice officials make a decision that has implications for public safety andcosts: Should a defendant be released pretrialor detained until adjudication? The period oftime between arrest and disposition (the pretrialstage) provides an opportunity to use advancements in science to safeguard the public.Bureau of Justice AssistanceU.S. Department of Justicecharge-based bail schedules have guidedcritical bail decisions. However, in the lastdecade, communities around the countryhave demonstrated that the decision torelease or detain a defendant pretrial canbe improved by assessing the defendants’risk. Pretrial instruments can help decisionmakers understand the likelihood of pretrialmisbehavior to improve public safety outcomesand save money.Historically, personal experience, professionaljudgment, confusing statutes, and monetaryChart 1: Data Driven – Pretrial Risk Assessment Instrument Accurately Identifies Pretrial Risks40%35%30%25%20%15%10%5%0%Low RiskModerate RiskFailure to Appear in CourtHigh RiskRearrest While on Pretrial ReleaseA validation study of the Kentucky Pretrial Risk Assessment Instrument demonstrates theinstrument’s ability to predict pretrial misconduct. As indicated in Chart 1, defendants categorizedas low risk have the lowest rates of pretrial failure, and those categorized as high risk have the highestrates (rates of pretrial failure are typically low, even among those labeled “high risk”).1ISSUE BRIEF1

1. What is a pretrial risk assessmentinstrument?tion is based on differences in statutes, data quality,availability of supervision resources, etc. It is important to validate any instrument on your populationand revalidate on a regular basis.A pretrial risk assessment instrument is typicallya one-page summary of the characteristics of anindividual that presents a score correspondingto his or her likelihood to fail to appear in courtor be rearrested prior to the completion of theircurrent case. Instruments typically consist of 7-10questions about the nature of the current offense,criminal history, and other stabilizing factors suchas employment, residency, drug use, and mentalhealth.23. Why is it important to knowsomeone’s risk level?A person’s risk level should be used to guide twodecisions: 1) the decision to release or detain pretrial, and 2) if released, the assignment of appropriate supervision conditions. While the chargefor which someone was arrested is important inthese decisions, the charge alone does not informyou of someone’s likelihood to make all court appearances or his or her risk to public safety.Responses to the questions are weighted, basedon data that shows how strongly each item isrelated to the risk of flight or rearrest duringpretrial release. Then the answers are tallied toproduce an overall risk score or level, which caninform the judge or other decisionmaker aboutthe best course of action.Research has shown that intensive supervisioncan work well with some higher-risk individuals,but it usually results in poor outcomes for peoplewho have been assessed to be lower risk.4The other reason to know someone’s risk scoreis to make the best use of scarce resources. It is awaste of money to over-condition people who donot need those conditions in order to comply. Itis a good use of money to provide supervision inthe community to someone who needs it, whencompared to the cost of housing, feeding, andproviding medical care in jail. Supervision cancost between 3- 6 a day. Jail (housing, feeding,medical care) can cost roughly 50 a day.2. What risk factors are on a pretrial riskassessment instrument?Unlike other risk/needs assessments, the pretrial risk assessment instrument contains factorsthat are associated with increased chances ofonly two types of failure during a short periodof time: failure to appear for all court hearingsand rearrest on a new charge. Typically, instruments weigh such factors as nature of the currentcharge, any pending charges, number of priorconvictions resulting in jail time, prior violentconvictions, failure to appear history, residentialstability, employment/caregiver history, and drugabuse history.3Most jurisdictions have bond schedules, whichbase a dollar amount set for pretrial release onthe charge for which the person was arrested.These schedules do not have any mechanism totake into account the individual risks posed bya defendant, and often they permit someone tobond out prior to being screened for risk.Which factors are predictive, and the weight of eachrisk factor, varies by jurisdiction. Some of this varia-A PUBLICATION OF THE PRETRIAL JUSTICE INSTITUTE2

4. How effective are pretrial riskassessment instruments?ment showed its efficacy by demonstrating thatlow-risk individuals had a 92.9% success rate,average-risk individuals had an 82.2% successrate, and high-risk individuals had a 68% successrate.6Pretrial risk assessment instruments are highlyeffective in their ability to predict rates of successon pretrial release (return to court and no newarrests while on release). Numerous studies haveshown that validated pretrial risk assessmentscan accurately differentiate defendants’ risk.5In the federal system, all 94 districts use a standardized instrument developed using their data. Chart2 shows the outcomes, with Risk Category 1 beingthe lowest risk and Risk Category 5 being thehighest. Those scored as Risk Category 5 still havereasonable outcomes when released pretrial withappropriate conditions of supervision.7The Virginia Risk Assessment Instrument usesa low/below average/average/above average/high scale to predict risk. A study of that instru-Chart 2: U.S. Pretrial and Probation, Federal Pretrial Risk Assessment(FTA Failure to Appear; NCA New Criminal Arrest; TV Technical TVRisk Category 1Risk Category 2Risk Category 4Risk Category 5ISSUE BRIEF3Risk Category 3

5. How are pretrial risk assessmentinstruments developed?6. What are the challenges andlimitations of pretrial riskassessment?Risk assessment instruments are developed bycollecting and analyzing local data to determinewhich factors, and to what degree, are predictiveof pretrial risks (failure to appear or rearrest).Careful consideration should be taken whendeveloping an instrument for your jurisdiction. There are a number of model instrumentsthat have been validated in one jurisdiction andadopted by other jurisdictions. In such cases,the jurisdiction collects data and then validatesthe imported tool on their own population. Allinstruments should be revalidated at regularintervals to ensure they retain their predictivevalidity. Differences in bail statutes, court rules,release options available to the court, and otherfactors cause variations from site to site. Statistical analysis on local data helps ensure these aretaken into account.Pretrial risk assessment instruments cannot predict with accuracy a specific individual’s behavior.The tools are research-based guides to takingrisks. There will be lower-risk individuals whofail on pretrial release, and there will be higherrisk individuals who succeed. However, theseinstruments provide an objective, standard wayof assessing the likelihood of pretrial failure thatresearch shows produces higher accuracy thansubjective assessments by even the most experienced and well-respected decisionmakers.This does not mean that pretrial risk assessmentinstruments should be used in place of professional discretion. The instrument produces ascore that can help anchor a decision, and deviations (overrides) can be expected and the reasonsshould be documented.There are a number of ways to accomplish thedevelopment of a locally valid instrument. Localuniversities and researchers may have the capacity to develop an instrument at reasonable cost.The U.S. Department of Justice may provideassistance under grant programs such as JusticeReinvestment or through their technical assistance programs.The pretrial risk assessment instrument is anappropriate tool for an in/out custody decisionbased on risk. Other instruments are available tohelp identify areas of need that should be addressed during supervision (such as the need formental health or substance abuse assessments).Finally, the instrument must be consistentlyvalidated to ensure its predictive validity. For thisto happen, jurisdictions must collect a variety ofdata about defendants, including if/how they arereleased pretrial, the quality and quantity of supervision conditions, their pretrial behavior, andthe adjudication of their case. This data can helpidentify any other predictive factors for pretrialsuccess, which programs can use to improve.A PUBLICATION OF THE PRETRIAL JUSTICE INSTITUTE4