Pretrial Risk Assessments Should Not Be Used, and If Used, Should Be Subject to Strict Safeguards.

Pretrial risk assessment should not be used, and if used, they should be subject to due process and other safeguards. Judges should be provided all of the information the risk assessments consider and the assessments should provide information as to risk but not recommend a specific bail and conditions of release.

In one jurisdiction, a police officer was killed by a suspect who had prior charges and convictions for assaulting public safety officers, but the justice of the peace did not have that information when he gave that defendant a personal recognizance bond. Comprehensive criminal history information, including the nature and circumstances of all charges and convictions, and all information pertaining to the statutory factors of bail considered by judges, should be provided to all judges at all levels of the judiciary in all jurisdictions within the United States of America.

Risk assessments instruments should no longer be used in setting bail or considering risk. To the extent such assessments are used, a party to a criminal case must decide to use such an assessment, and a criminal defendant must agree to be assessed.. The proponent of the assessment and the result must defend the validity and calibration of such assessment if such becomes an issue in a criminal matter.

Risk assessments or risk assessment processes should not make specific recommendations as the the bail or specific conditions of release as part of a decision making framework, because that is not scientific or evidence-based. The relative risk class of the defendant, e.g., 1, 2, 3, etc., and what those categories mean, e.g., category risk four represents a 34% failure to appear rate and a 15% risk of committing a new offense, should be the information that is presented to judges.

Risk assessments are not individualized consideration. They are instead dehumanizing because they label people with numbers and sort them into arbitrary categories calibrated not by science but by setting a substantive tolerance of risk. The assessments are cloaked in science, yet they are calibrated by people who are inherently biased and they are set by ex parte agreements among local counsels of several branches of government often behind closed doors. By employing an algorithm, we lock in the past, prevent change, and guarantee that all institutional biases from which we work to reduce going forward are not reduced because they are locked in by assumptions made by the builders of a computer program.

In addition, risk assessments fail to, in an evidence-based fashion, provide judges any information on risk reduction and how to address criminogenic needs when offenders cannot be constitutionally ordered to treatment prior to conviction, there are little services available to them other than supervision and monitoring by the state, and thus there is virtually little hope that any reduction in crime could occur. The assessments tell judges who is risky, but they don’t tell judges how to reduce that risk other than to expand the dragnet of pre-conviction monitoring and supervision.

Risk assessments have done nothing over a generation to reduce mass incarceration, and instead have been blamed for over-labeling people, particularly racial minorities, as risky to further justify the expansion of the penal state, prevent true rehabilitation, and lock out the possibility of healing and remorse by permanently affixing a criminal risk scarlet letter onto a person. America can do better.