WHEREAS, the sentencing proceeding is one of the most important stages of a criminal case and Due Process requires that every sentence be based on highly individualized factors including the specific circumstances of the crime and the specific history and characteristics of the defendant being sentenced;

WHEREAS, there is a current trend in criminal justice policy to develop predictive assessment tools (often called risk assessment tools) to guide or inform decisions about pretrial detention, prison classification, sentencing, and parole and probation supervision;

WHEREAS, developers of risk assessment tools conduct a statistical analysis of large, aggregated data sets of criminal behavior over time, attempt to identify “risk factors” that correlate with future criminal conduct and then develop an algorithm that calculates an individual’s statistical likelihood of future criminal conduct on the basis of the number of risk factors that apply;

WHEREAS, such risk factors, which often include age, gender, past criminal history, education level, marital status, and employment status, can serve to exacerbate racial disparity and punish poor and minority defendants based not on their actual acts but on their socio-demographic characteristics;

WHEREAS, any classification that includes age and gender raises Equal Protection concerns;

WHEREAS, during the sentencing process in particular, Due Process is offended when an individual defendant’s term of imprisonment is based on possible future misconduct, as extrapolated from other people’s past behavior;

WHEREAS, individuals should be treated as individuals regardless of the accuracy of generalizations about others in their group;

WHEREAS, in 2010, the Pennsylvania Risk Assessment Law mandated the Pennsylvania Commission on Sentencing (“Sentencing Commission”) to develop a risk assessment tool to be used by judges at the sentencing stage of a criminal proceeding;

WHEREAS, in Spring 2018, the Sentencing Commission released for public comment a proposed risk assessment tool which categorizes defendants as “typical risk offender,” “high risk offender,” and “low risk offender” using as risk factors age, gender, and past criminal history;1

WHEREAS, the tool proposed by the Sentencing Commission over-predicts the risk of future criminal conduct because, in designing the algorithm, recidivism was defined to include mere arrests and technical violations of parole or probation conditions as opposed to actual criminal convictions;2

WHEREAS, mere arrests and technical violations of probation and/or parole correlate more to neighborhood and geographical areas of police activity than to future dangerousness;

WHEREAS, the Sentencing Commission self-reports that the current proposed risk assessment tool has “accuracy rates of around 60%-85%, depending on the measure used and the risk scale in question;”3

WHEREAS, the Sentencing Commission also self-reports that the proposed tool results in high risk predictions for “a crime against a person” which does “not meet an acceptable level of predictive accuracy;”4

WHEREAS, these accuracy rates, and the vagueness with which they are reported, are unacceptable: an accuracy rate of 60% indicates that the tool would be wrong in 40 out of 100 cases, and an accuracy rate of 50% is the same as a coin toss;

WHEREAS, the Sentencing Commission reports using risk development data from 2004 to 2006 in which black offenders had base recidivism rates that averaged 11% higher than the base recidivism rates of white offenders;5

WHEREAS, This is inconsistent with a March 2018 report from the Philadelphia Reentry Coalition which, in breaking down recidivism rates by race in Philadelphia, found that black offenders had the lowest recidivism rate at 32.5%, followed by Hispanics at 35.7%, and whites at 39.8%;6

WHEREAS, the difference in these reports raises the question of whether there is an inherent racial bias in the data set used by the Sentencing Commission;

WHEREAS, categorizing a defendant as typical, high or low risk on the basis of a “scientific” tool will serve to discourage judicial discretion and judicial consideration of individualized mitigating and aggravating factors such as a defendant’s family support; a defendant’s drug, alcohol, and mental health characteristics; comments by the victims of the offense; and other circumstances and factors specific to the defendant and to the offense of conviction;

NOW, THEREFORE, BE IT RESOLVED that the Philadelphia Bar Association opposes the implementation or utilization of risk assessment tools at time of sentencing. Such tools are contrary to the requirement of individualized sentencing and the requirement that individuals be punished on the basis of their own unique circumstances rather than on the basis of predicted behavior extracted from the history of other people. Such tools also have questionable predictive accuracy and can serve to exacerbate racial disparities already present in the criminal justice system. Moreover, the particular tool released in Spring 2018 by the Sentencing Commission violates Equal Protection and Due Process principles and allows for sentences to be imposed on the basis of crimes that have yet to be, and may never be, committed.

AND BE IT FURTHER RESOLVED that the Board of Governors authorizes the Chancellor or the Chancellor’s designee to communicate the position of the Philadelphia Bar Association to the Pennsylvania Commission on Sentencing, the Supreme Court of Pennsylvania, the Governor, the General Assembly, members of the legal profession and the community, and to take whatever action may be appropriate to effectuate this resolution.

ADOPTED: May 30, 2018

1 Pennsylvania Commission on Sentencing, “Proposed Sentence Risk Assessment Instrument” at 10 (March 2018), available at

2Id. at 9.

3Id. at 4.


5Pennsylvania Commission on Sentencing, “Risk Assessment Update” at 2 (Nov. 2017), available at visited May 8, 2018).

6The Philadelphia Reentry Coalition, Data and Metric Subcommittee, “Calculating a Unified Recidivism Rate for Philadelphia: A Data Snapshot of Reentry and Recidivism 2012-2015” at 13 (March 2018), available at (last visited May 8, 2018).