In an exclusive interview granted to Time Magazine, Attorney General Eric Holder expressed strong concerns about the equity of empirical risk assessments used to determine how a sentence will be carried out.  His concern is that “static” risk factors (those that are largely unchangeable through rehabilitation like educational attainment and employment history) unduly influence these assessment will lead to young, black males being unfairly sentenced in comparison to more socially successful white collar criminals.

Mr. Holder’s suspicion that empirical risk assessments may exacerbate (or, to give a charitable reading, perpetuate) racial inequality in the criminal justice system is not entirely without merit.  Earlier risk assessment instruments relied heavily on static factors; those which are immutable, unmalleable, and in sum less predictive than those that can be addressed.  Items like criminal history have been suggested to be emblematic of distal racial bias, though do not carry enough weight in the calculation of today’s third and fourth generation for this to be so.

What has been shown to predict recidivism are elements that are largely changeable, such as associating with criminal peers or harboring antisocial thoughts.  These items are more likely to capture the risk of an individual reoffending.  Race, on the other hand, has almost no independent predictive validity.

Perhaps Mr. Holder’s most puzzling comment is of how these risk factors influence sentencing.  Perhaps he’s referring to the body of research that suggests that an individual detained before the trial phase are more likely to be found guilty and receive a custodial sanction, though this has been shown to have little correlation with race.  The adoption of these instruments allows low risk offenders of ALL races and ethnicities to be monitored in the community sooner.

There is little substance or logic to the concerns about “data-driven sentencing” expressed by the sitting Attorney General.  Actuarial risk does not determine guilt.  Actuarial risk does not determine the sentence.  Actuarial risk is used to determine what the most effective, cost-efficient modality for the offender to complete his or her sentence under at the least cost to public safety.

Risk assessment instruments rout out the specter of latent biases in lieu of actuarial probability.  What little interrelation there may be between criminality, risk factors, and race independently is spurious and pales in comparison to the manifold benefit these tools afford those who use them.