THE GENERAL ASSEMBLY
Title 204—JUDICIAL SYSTEM GENERAL PROVISIONS
COMMISSION ON SENTENCING
PART VIII. CRIMINAL SENTENCING
[ 204 PA. CODE CH. 305 ]
Proposed Sentence Risk Assessment Instrument
[49 Pa.B. 3718]
[Saturday, July 20, 2019]The Pennsylvania Commission on Sentencing (Commission) hereby publishes for public comment a proposed Sentence Risk Assessment Instrument, 204 Pa. Code §§ 305.1—305.9, for use by the sentencing court to help determine the appropriate sentence within the limits established by law. This is the fifth proposal of a Sentence Risk Assessment Instrument published by the Commission. The four previous proposals, published in the Pennsylvania Bulletin, are listed as follows:
• 45 Pa.B. 1751 (April 11, 2015)
• 47 Pa.B. 1999 (April 8, 2017)
• 48 Pa.B. 2367 (April 28, 2018)
• 48 Pa.B. 6961 (November 3, 2018)
Act 2010-95 mandates the Commission to ''. . . adopt a Sentence Risk Assessment Instrument for the sentencing court to use to help determine the appropriate sentence within the limits established by law. . .The risk assessment instrument may be used as an aide in evaluating the relative risk that an offender will reoffend and be a threat to public safety'' (42 Pa.C.S. § 2154.7). In addition to considering the risk of re-offense and threat to public safety, Act 2010-95 also permits the risk assessment instrument to be used to determine whether a more thorough assessment is necessary, or as an aid in determining appropriate candidates for alternative sentencing (e.g., County Intermediate Punishment, State Intermediate Punishment, State Motivational Boot Camp, and Recidivism Risk Reduction Incentive).
The current proposed Sentence Risk Assessment Instrument, approved by the Commission on June 13, 2019, for public comment, is set forth in Annex A. The proposal includes the following modifications to the November 3, 2018, proposal:
• Eliminate the 'Crime Against a Person' scale.
• Retain the general risk scale but adjust the 'high risk' threshold to improve accuracy.
• Eliminate the 'Sentence Risk Assessment Summary' and all risk labels and replace with a notation of 'RNR Report Recommended' on the Guideline Sentence Form for any offender other than a 'typical risk offender.'
In addition, the Commission is requesting comment on two implementation issues:
• Scope—if adopted, should the initial implementation of the Sentence Risk Assessment Instrument apply to all non-DUI cases, or should it be limited to certain types of cases (e.g., most serious offense of the judicial proceeding is a felony or M1)?
• Funding—if adopted, should the implementation of the Sentence Risk Assessment Instrument be linked to funding for preparation of RNR Reports?
In accordance with 42 Pa.C.S. § 2155, the Commission shall publish in the Pennsylvania Bulletin all proposed sentencing guidelines and the Sentence Risk Assessment Instrument and hold public hearings not earlier than 30 days and not later than 60 days thereafter to afford an opportunity for the following persons and organizations to testify:
(i) Pennsylvania District Attorneys Association
(ii) Chiefs of Police Associations
(iii) Fraternal Order of Police
(iv) Public Defenders Organization
(v) Law School faculty members
(vi) Pennsylvania Board of Probation and Parole
(vii) Pennsylvania Department of Corrections
(viii) Pennsylvania Bar Association
(ix) Pennsylvania Wardens Association
(x) Pennsylvania Association on Probation, Parole and Corrections
(xi) Pennsylvania Conference of State Trial Judges
(xii) Any other interested person or organization
The Commission will hold the following three public hearings to receive comments on the proposed Sentence Risk Assessment Instrument:
Hearing I.
Date: Tuesday, August 20, 2019
Time: 1:30 p.m.
Location: Juanita Kidd Stout Center for Criminal Jus- tice, Courtroom 304
1301 Filbert Street
Philadelphia, PA 19107Hearing II.
Date: Wednesday, August 21, 2019
Time: 10:00 a.m.
Location: Pennsylvania Judicial Center
Conference Room A
601 Commonwealth Avenue
Harrisburg, PA 17120Hearing III.
Date: Thursday, August 22, 2019
Time: 1:30 p.m.
Location: Allegheny County Courthouse
Courtroom 533
436 Grant Street
Pittsburgh, PA 15219Persons or organizations wishing to testify are asked to contact the Commission at least five business days prior to the hearing to be registered and are asked to provide an electronic copy of any testimony when registering. Written comments from persons or organizations not wishing to testify should be received by the Commission at least five business days before the last-scheduled public hearing. Commission staff will copy and distribute to Commission members any documents submitted at least five business days prior to the hearing. Contact Cathy Dittman (814.863.5729 or CWD2@PSU.EDU) to register to testify and to submit testimony or written comments.
Any other individuals wishing to testify who have not registered at least 5 business days prior to the hearing are asked to provide 30 copies of any testimony at the time of the public hearing. All Commission meetings and hearings are open to the public.
The Commission will evaluate the proposed Sentence Risk Assessment Instrument set forth in Annex A after consideration of the testimony and written comments received. Any instrument adopted by the Commission is submitted to the General Assembly for review by way of publication in the Pennsylvania Bulletin, and will become effective 90 days after publication unless rejected by concurrent resolution of the General Assembly.
JUDGE SHEILA A. WOODS-SKIPPER,
Chair
Commentary on Annex A This Commentary addresses issues considered and modifications approved by the Commission relating to the proposed Sentence Risk Assessment Instrument published in the Pennsylvania Bulletin on November 3, 2018. Included in the review by the Commission were: testimony and written comments received during five public hearings held in December 2018; an external review of the proposed Sentence Risk Assessment Instrument conducted by a Heinz College Systems Synthesis Project Team at Carnegie Mellon University (CMU); and a Recommendation and Report submitted by the Pennsylvania Bar Association (PBA). Copies of testimony received and other documents are posted on the Commission's web site (http://pasentencing.us). Information on the CMU external review and the recommendation of the PBA are discussed as follows.
The current proposed Sentence Risk Assessment Instrument, approved by the Commission on June 13, 2019, for public comment, is set forth in Annex A. The proposal includes the following modifications to the November 3, 2018, proposal:
• Eliminate the 'Crime Against a Person' scale;
• Retain the general risk scale but adjust the 'high risk' threshold to improve accuracy; and
• Eliminate the 'Sentence Risk Assessment Summary' and all risk labels and replace with a notation of 'RNR Report Recommended' on the Guideline Sentence Form for any offender other than a 'typical risk offender.'
CMU External Review
In December 2018, Commission staff submitted a Systems Project proposal to CMU. The proposal called for a team of CMU Heinz College Master's students to conduct an external review of the proposed sentence risk assessment instrument. The proposal was accepted by CMU, and Commission staff provided the research team with a dataset of offenders sentenced during 2004—2006 and released within a 3-year follow-up period. The CMU team was asked to validate the Commission's prior analyses, evaluate the proposed sentence risk assessment instrument for validity, accuracy, and fairness, and suggest improvements to the design of the instrument. A final presentation and report were received in May 2019. The report included four CMU recommendations followed by Commission responses.
CMU Recommendation 1: Restrict the usage of the general recidivism risk scale to low risk.
CMU students concluded that the general recidivism risk scale performed moderately well and compared favorably to other well-known widely accepted risk assessments, using the Area Under the Curve (AUC) measure. However, when these measures of performance were broken out by low and high risk groups, they find that the high risk predictions are less accurate (high risk error rate = .48) than the low risk predictions (low risk error rate = .16). This suggests that for high risk predictions, the model misclassifies offenders 48% of the time (48% do not recidivate). Conversely, for low risk predictions, only 16% are misclassified (16% recidivate). Or stated another way, 84% of offenders classified as low risk do not recidivate. The CMU team concluded that the high risk error rate was too high and that the tool should only make use of the low risk group. However, the CMU team and Commission staff both note that the higher error rate in the high risk predictions can be explained by the fact that fewer people recidivate than do not recidivate (33% vs. 67%). By chance, a prediction of low risk is more likely to be correct than a high risk prediction.
While the high risk error rate, as reported by the CMU team, is higher than the low risk error rate, the Commission feels that CMU's analysis is incomplete. Their focus is solely on the accuracy within the predicted outcome (percent of high (low) predictions that recidivate (not recidivate)) as opposed to the accuracy within the actual outcome (percent of those who recidivate (not recidivate) that are predicted as high (low) risk). In response, the Commission asserts that both measures of accuracy need to be considered.
An examination of the accuracy within the actual outcome produces a different picture than just looking at the accuracy within the predicted outcome. Of the people who did not recidivate, the instrument correctly predicted 64% as low risk, and of the people that did recidivate, the instrument correctly predicted 77% as high risk. These are much high values that debunk the myth that the instrument performs no better than a coin flip. An overview of the values of accuracy are provided as follows in Exhibit 1.
Exhibit 1: Overview of Values of Accuracy
No Recidivate Total
--------------------------------------
Low Risk 19,431 3,507 22,938
High Risk 10,967 11,947 22,914
------------------------------------------------------------------
Total 30,398 15,454 45,852• Accuracy within the predicted outcome for low risk is 85% (19,431 ÷ 22,938).
• Accuracy within the predicted outcome for high risk is 52% (11,947 ÷ 22,914).
• Accuracy within the actual outcome for low risk is 64% (19,431 ÷ 30,398).
• Accuracy within the actual outcome for high risk is 77% (11,947 ÷ 15,454).
In conclusion, the Commission argues that both the low and high risk predictions should be utilized, especially in light of reconsideration of the way information will be shared with judges. The Commission is sensitive to concerns of labeling an offender as high risk. As such, the Commission also recommends that the sentencing guideline form only display ''RNR Report Recommended'' when the offender is classified as low or high risk. Within this framework, a risk summary form containing detailed information on the risk factors, points assigned, and risk category will no longer be prepared.
CMU Recommendation 2: Do not use the Crime Against a Person risk scale—not even to identify low risk.
The Commission agrees with the recommendation to drop the Crime Against a Person risk scale. Earlier analyses revealed that Crimes Against a Person are a rare event (slightly less than 5% recidivate) and that predictions of risk generated an unacceptable level of false positives. This led the Commission to vote to drop the high risk component of the scale.
On the low risk side, the Commission previously argued that providing judges with additional information when an offender was low risk of a Crime Against and Person and high risk of general recidivism could potentially mitigate the high-risk label and help the judge determine a more appropriate sentence. However, since the Commission is considering not including the low and high risk labels on the sentencing guidelines form, these cases will already be flagged for the judge to receive additional information.
CMU Recommendation 3: Move the high risk cut point for the general recidivism scale from 10 to 18, to 12 to 18.
The Commission agrees that moving the risk cut points for the high-risk group would increase the accuracy of the high-risk group (accuracy of the prediction). However, there are two significant policy implications associated with this decision. First, the number of individuals classified as high risk declines from roughly 16% (one standard deviation) to slightly less than 5% of offenders. This significantly reduces the number of offenders recommended to receive additional information.
Second, maintaining a statistical rationale for the selection of low and high risk cut points is desirable. The mean risk score plus and minus one standard deviation addresses the concept of ''relative risk that an offender will reoffend and be a threat to public safety'' that is outlined in the Commission's mandate (42 Pa.C.S. § 2154.7). A standardized method to assign cut points avoids the political/unsystematic process of picking and choosing cut points that consider the ratio of false positives to false negatives. For precisely this reason the Commission previously voted to move away from this method in favor of the mean and standard deviation approach. The latter approach identifies approximately 16% of offenders as low risk, 68% of offender as typical risk and 16% of offenders as high risk, which greatly increase the practical utility of the risk assessment instrument. This is especially salient now that the Commission is considering removing any indication of low and high risk labels on the sentence guideline form.
Notwithstanding these policy concerns, the Commission voted to increase the threshold for high risk from 10 to 12, thus improving the accuracy of the instrument.
CMU Recommendation 4: Remove gender from the general recidivism risk scales.
Because of constitutional concerns raised during the public hearing process, Commission staff specifically asked CMU to investigate the impact of removing gender from the instrument. A previous memo by Commission staff (December 2018) showed that removing gender has a marginal impact on the accuracy of the sentence risk assessment instrument. Overall, the accuracy rate is still moderate to strong and the accuracy of low and high risk predictions remain essentially the same. However, removing gender results in fewer females classified as low risk and more females classified as high risk.
Three options regarding gender and risk
Model 1 [Current]: Gender is considered a risk factor and one point is assigned to males
In the current sentence risk assessment instrument gender is a scored risk factor (Female=0; Male=1). The current risk scale is 18 points and the cut points for typical risk are 5 through 9. Considering gender as a risk factor enables the model to account for the significant difference in recidivism rates between males (35%) and females (26%). Due to this difference in recidivism rates between males and females, the model assigns more females into the low risk category (30.5%), and less females into the high risk category (7.9%). In comparison, males are split evenly at roughly 16% between low and high risk.
Model 2: Do not assign points to gender (males and females given score of zero)
Removing gender as a risk factor results in both females and males receiving zero points (Female = 0; Male = 0). Not assigning points to gender decreases the risk scale by 1 point and shifts the cut points for typical risk from 5 through 9 to 4 through 8. Removing points for gender has the greatest impact on females. Because of the shift in the cut points, fewer females are classified as low risk (17.3%) and more females are classified high risk (14.0%). There is no impact on males due to the removal of 1 risk point and the shifting of the cut points by 1 point. The overall accuracy of the model, for males and females combined, remains the same due to the low number of females (20%); however, this model decreases the accuracy of the high risk prediction for females because the model is assigning more females to high risk that do not recidivate.
Model 3: Create separate risk scales for males and females
Since males make up 80% of the overall sample, there are very few implications for males when a male only scale is created. The risk scale goes from 0 to 18 to 0 to 17 and the cut points shift to 4 through 8 for typical risk (point for gender is removed). In contrast, the implications for females is greater. The risk scale shifts to 0 to 12 and the cut points for typical risk are 2 through 5. Under this model, males have roughly 16% in low and high risk, while 9.8% of females are classified as low risk and 17.8% are classified as high. As a result, the measures of overall accuracy and the percentage predicted high risk who recidivate declines.
A move away from the current model, either through removing a point for males or creating separate scales, leads to more females being classified as high risk. This leads to a reduction in the overall percentage of females correctly classified (78% to 61%) and reduces the positive predicted value (PPV) from 50% to 45%.
Based on these findings, the Commission voted to retain gender in the instrument. This was based in great part on the determination that males would be unaffected by including consideration of gender, but that females would be negatively impacted (reduced accuracy) if gender was not considered.
PBA Recommendation
The PBA's House of Delegates voted unanimously to recommend that the Commission reject its proposed Risk Assessment Instrument. However, the PBA suggested that the Instrument could be re-oriented to obtain beneficial results:
If the goal of a risk assessment instrument is to provide more information to sentencing judges through the creation of pre-sentence reports, then there is no need for the judges to actually see the risk assessments. With an amendment to Pa.R.Crim.P 702, county probation offices could prepare and provide pre-sentence reports sua sponte. And, the parties could be prohibited from disclosing the results of the risk assessment to the sentencing judge. True, judges might have an inkling that a defendant scored high or low on the Instrument, but the lack of a concrete number would be less prejudicial to defendants.The Commission's approach throughout the development of the sentence risk assessment instrument has been complete transparency, illustrated by the previously proposed 'Sentence Risk Assessment Summary' which included a listing of all factors used to determine the risk score of the offender, details of all the recidivism rates by risk score, identification of risk categories, and information about the accuracy of the instrument. The recommendation proposed by the PBA, intended to address concerns associated with the labeling of an offender based on a risk category, is to restrict the availability of this information and eliminate the use of the risk category labels.
The Commission voted to approve this recommendation from the PBA and to modify the proposal to eliminate the dissemination of information related to risk. In place of the 'Sentence Risk Assessment Summary,' the Commission's Guideline Sentence Form will include the notation 'RNR Report Recommended' in cases where the risk score is higher or lower than 'typical risk.'
Proposed Utilization
The RNR Model (risk, needs, responsivity) suggests that consideration of both risk and needs is necessary in order to determine the most appropriate sentence; risk relates to level of supervision, and needs informs the type and intensity of treatment. Programs such as state intermediate punishment, certain county intermediate punishment, and problem solving courts are most effective for high risk/high needs offenders, since these programs provide both supervision and treatment. And low risk/low needs offenders may be candidates for diversion. While some argue that the risk assessment instrument considers many of the same factors considered in the sentencing guidelines, the analysis finds that for every cell (OGS/PRS combination) of the sentencing guidelines, both high risk and low risk cases are present. So while the sentencing guidelines are based on retribution and recommend a range of sanctions available to the court, RNR can assist the court in determining the choice among those options.
Regarding resource utilization and funding, recent surveys of county adult probation and parole offices found that all but two counties were using a risk/needs assessment, that the assessments were most often used to determine probation and county parole caseloads, and that many offices conducted reassessments.
Based on the Commission's analysis of its sample, the proposal recommends assessments of about 9,750 high or low risk non-DUI cases statewide each year, of which approximately 1,000 cases resulted in sentences to state prison. For those counties conducting assessments of probation and county parole cases, the ability to shift the preparation of these probation and county parole assessments from post-sentence to pre-sentence would limit the impact of the proposal to an 11% increase in assessments.
Annex A
TITLE 204. JUDICIAL SYSTEM GENERAL PROVISIONS
PART VIII. CRIMINAL SENTENCING
CHAPTER 305. SENTENCE RISK ASSESSMENT INSTRUMENT Sec.
305.1. Preliminary provisions. 305.2. Sentence Risk Assessment Instrument methodology. 305.3. Sentence Risk Assessment Instrument standards. 305.4. Sentence Risk Assessment Instrument procedures. 305.5. Sentence Risk Assessment Instrument recommendation—general. 305.6. Offense Types. Classification of Prior Convictions and Current Convictions, including Inchoates. 305.7. Recidivism Risk Scales. 305.8. Recidivism Rates by Risk Score. 305.9. Guideline Sentence Form. § 305.1. Preliminary provisions.
(a) Authorization.
(1) As authorized by 42 Pa.C.S. § 2154.7 (relating to adoption of risk assessment instrument), the Commission shall adopt a Sentence Risk Assessment Instrument for the sentencing court to use to help determine the appropriate sentence within the limits established by law for defendants who plead guilty or nolo contendere to, or who were found guilty of, felonies and misdemeanors.
(2) The Sentence Risk Assessment Instrument may be incorporated into the sentencing guidelines under 42 Pa.C.S. § 2154 (relating to adoption of sentencing guidelines).
(3) The Sentence Risk Assessment Instrument may be used to determine whether a more thorough assessment is necessary and to order a presentence investigation report.
(4) The Sentence Risk Assessment Instrument may be an aid to help determine appropriate candidates for alternative sentencing.
(b) Definitions. For the purposes of this chapter:
(1) ''Actuarial risk assessment.'' A statistical method of estimating the risk of recidivism; the probability of recidivating is related to the proportion of offenders in a risk category who recidivate.
(2) ''Common Pleas Case Management System (CPCMS).'' A web-based application operated by the Administrative Office of Pennsylvania Courts which serves as the source of data for determining number of prior convictions and associated offense type.
(3) ''Conviction.'' An offense for which a defendant pleads guilty or nolo contendere, is found guilty following a bench or jury trial, or is accepted for participation in an authorized diversion program, including Accelerated Rehabilitative Disposition (relating to Pa.R.Crim.Proc. Chapter 3), Probation without Verdict (relating to 35 P.S. § 780-117) or Disposition in Lieu of Trial or Criminal Punishment (relating to 35 P.S. § 780-118).
(4) ''DUI offender.'' An offender for whom the most serious offense of the judicial proceeding is DUI. The Sentence Risk Assessment Instrument does not apply if DUI is the most serious offense of the judicial proceeding. The Sentence Risk Assessment Instrument does apply if DUI is an offense other than the most serious offense in the judicial proceeding.
(5) ''Judicial proceeding.'' A sentencing hearing in which all offenses for which the offender is convicted are pending before the court for sentencing at the same time. A judicial proceeding may include multiple OTNs.
(6) ''Magisterial District Judge System (MDJS).'' A web-based application operated by the Administrative Office of Pennsylvania Courts which serves as the source of data for determining the number of prior convictions and associated offense type following a final disposition by a minor court.
(7) ''Minor courts.'' A court of limited jurisdiction with authority to preside at preliminary arraignments and preliminary hearings, dismiss complaints, conduct trials and/or accept guilty pleas for misdemeanors, and hold cases for trial in the courts of common pleas. This includes Philadelphia Municipal Court and Magisterial District Judges.
(8) ''Offense gravity score (OGS).'' An assignment in the sentencing guidelines reflecting the seriousness of a conviction offense. The OGS assigned to the most serious offense in the judicial proceeding (and if more than one offense, then the offense with the highest OGS, longest statutory maximum, and then the offense entered first in SGS Web) is used to determine the risk factors and associated values to be included in the risk scales.
(9) ''Offense tracking number (OTN).'' A unique identifying number assigned to an entire set of charges related to a conviction. An OTN is generally assigned by the court at the time of arraignment.
(10) ''Offense types.'' The classification of prior convictions and current convictions, including inchoates, as provided follows:
i. Murder;
ii. Danger to person:
1. felonies;
2. misdemeanors;iii. Sexual:
1. felonies;
2. misdemeanors;iv. Burglary;
v. Property:
1. felonies;
2. misdemeanors;vi. Public administration;
vii. Public order;
viii. Firearms;
ix. Other weapons;
x. Drug:
1. felonies;
2. misdemeanors;xi. DUI;
xii. Other.
A detailed description of the offense types is located at § 305.6. Any unlisted offense, or any new or amended offense, will be assigned an offense type by the Commission based on the current equivalent offense type.
(11) ''Pre-sentence investigation (PSI) report.'' A report, authorized by the Rules of Criminal Procedure (Pa.R.Crim.P. Rules 702-703), that includes information regarding the circumstances of the offense and the character of the defendant sufficient to assist the judge in determining sentence.
(12) ''Recidivism.'' A re-offense for any crime, defined as a re-arrest for a felony or misdemeanor in Pennsylvania within three years of the imposition of a sentence to the community or within three years of release from confinement resulting in a conviction. Re-offense does not include out-of-state, federal or foreign charges or technical violations.
(13) ''Risk assessment instrument.'' An empirically based worksheet which uses factors that are relevant in predicting recidivism. A risk assessment instrument is often identified with a specific 'generation' of development:
i. 1st generation. ''Professional judgment,'' with assessments based on training and experience;
ii. 2nd generation. ''Risk assessment,'' an actuarial assessment of static criminal justice and demographic factors used to estimate risk of re-offense;
iii. 3rd generation. ''Risk-needs assessment (RNA),'' an actuarial assessment of static and dynamic factors and changing circumstances, such as relationships, employment, and substance abuse, used to estimate risk of re-offense and to assess criminogenic needs to be addressed through treatment and supervision;
iv. 4th generation. ''Risk-needs-responsivity (RNR) assessment,'' an actuarial assessment of static and dynamic factors used to match the level of service to the offender's risk to re-offend; assess criminogenic needs and target them in treatment; and structure the sentence to address the learning style, motivation, abilities and strengths of the offender.
(14) ''Risk category.'' The relative risk of recidivism as compared to other, based on the Sentence Risk Assessment Instrument:
i. ''Typical risk offender.'' An offender with a risk score of 5 points or greater but less than 12 points who shares characteristics with offenders at typical risk of recidivism. This middle category contains approximately 78% of offenders.
ii. ''Low risk offender.'' An offender with a risk score less than 5 points who shares characteristics with offenders at lower risk of recidivism. This low category contains approximately 18% of offenders.
iii. ''High risk offender.'' An offender with a risk score of 12 points or greater who shares characteristics with offenders at higher risk of recidivism. This high category contains approximately 4% of offenders.
(15) ''Risk factors.'' The variables obtained from AOPC or Commission data and determined to be statistically significant in relation to the risk of recidivism. Risk factors selected by the Commission and included in the Sentence Risk Assessment Instrument include:
i. ''Age.'' Based on Commission data, the age of the offender calculated using date of birth as reported on guideline sentence form and the date the Sentence Risk Assessment Summary is generated.
ii. ''Gender.'' Based on Commission data, the sex of the offender reported on the guideline sentence form.
iii. ''Number of prior convictions.'' Based on prior conviction data provided by AOPC, the number of unique OTNs associated with Pennsylvania convictions occurring before the date the Sentence Risk Assessment Summary is generated.
iv. ''Prior conviction offense type.'' Based on prior conviction data provided by AOPC, all offense types identified for any felony or misdemeanor convictions.
v. ''Current conviction offense type.'' Based on Commission data, the offense type of the most serious offense of the judicial proceeding.
vi. ''Multiple current convictions.'' Based on Commission data, more than one current conviction offense in the current judicial proceeding.
vii. ''Prior juvenile adjudication.'' Based on Commission data, any juvenile adjudication included in the prior record score for the judicial proceeding is used in the risk scales. This may include prior juvenile adjudications that lapse for purposes of the calculation of the Prior Record Score.
(16) ''Risk scale.'' Measures of the outcomes derived from statistical models used to determine the relative risk to recidivate based on identified factors.
(17) ''RNR Report.'' A report based on a risk-needs-responsivity assessment of static and dynamic factors used to match the level of service to the offender's risk to re-offend; assessing criminogenic needs and targeting them in treatment; and containing recommendations regarding the structuring of the sentence to address the learning style, motivation, abilities and strengths of the offender.
(18) ''Sentencing Guidelines Software Web Application (SGS Web).'' A JNET-based application operated by the Commission which includes the modules for Sentencing Guidelines and for the Sentence Risk Assessment Instrument. SGS Web serves as the source of data for determining an offender's age, gender, current conviction offense type, multiple current convictions and prior juvenile adjudications, as well as determining the most serious offense of a judicial proceeding.
(19) ''Sentence Risk Assessment Instrument.'' The actuarial tool, adopted by the Commission and deployed through SGS Web, used to identify cases for which an RNR Report Recommendation is generated. The Instrument uses factors that are relevant in predicting recidivism to estimate risk.
§ 305.2. Sentence Risk Assessment Instrument methodology.
(a) Development and validation.
(1) The Commission conducted a study using offenders sentenced during 2004—2006 (n=131,055) to allow for a three-year follow-up period for all offenders, while taking advantage of improvements in data quality and a more recent time period. Additional information on the development and validation of the Sentence Risk Assessment Instrument is available from the Commission.
(2) The Commission developed risk scales for all offenders, except those with DUI as the most serious offense of the judicial proceeding.
(3) In developing the risk scales, the following analyses were conducted:
i. bivariate analyses to determine which factors were related to recidivism;
ii. multivariate logistic regression to determine which factors best predicted recidivism while holding other factors constant;
iii. rotation of all categories for factors that were multi-categorical to ensure that reported differences were real and not due to a particular comparison category;
iv. Receiver Operating Characteristic (ROC) analysis, which plots the true positive rate (i.e., how many people were predicted to recidivate and did recidivate) against the false positive rate (i.e., how many people were predicted to recidivate but did not recidivate); and
v. validation of the final scales with both samples.
(b) Risk factors and scales—general.
(1) Based upon the analyses conducted by the Commission, the following factors were found to be predictive of recidivism, and thus, used in the risk assessment scales:
i. age;
ii. gender;
iii. number of prior convictions;
iv. prior conviction offense type;
v. current conviction offense type;
vi. multiple current convictions;
vii. prior juvenile adjudication.
(2) The risk scale for recidivism is located at § 305.7.
(c) Recidivism rates—general.
(1) The Sentence Risk Assessment Instrument provides the probability of recidivism based on the proportion of offenders in the development and validation samples who recidivate. Offenders identified as low risk or high risk were found to be significantly different in risk of recidivism than the typical offender.
(2) Recidivism rates by risk score for recidivism are located at § 305.8.
§ 305.3. Sentence Risk Assessment Instrument standards.
(a) Effective January 1, 2020, the court shall use the SGS Web-based Sentence Risk Assessment Instrument to identify cases for which an RNR Report is recommended to help determine the appropriate sentence for non-DUI offenders who plead guilty or nolo contendere to or who were found guilty of felonies or misdemeanors.
(b) The Sentence Risk Assessment Instrument does not apply to sentences imposed as a result of the following: accelerated rehabilitative disposition; disposition in lieu of trial; direct or indirect contempt of court; violations of protection from abuse orders; negotiated pleas; or revocation of probation, county intermediate punishment, state intermediate punishment or parole.
(c) In every case in which an RNR Report is recommended and submitted, the RNR Report shall be considered by the court prior to sentencing. The court is encouraged to determine procedures for the preparation of RNR Reports.
(d) In every case in which a court of record imposes a sentence for a felony or misdemeanor, and an RNR Report is recommended, the court shall make as a part of the record, and shall record on the guideline sentence form whether the court received and considered an RNR Report. This information shall be electronically transmitted to the Pennsylvania Commission on Sentencing in the manner described in § 303.1(e).
(e) Unless otherwise provided by the Commission, the JNET-based Sentencing Guidelines Software Web application (SGS Web) shall be used at the court's direction to prepare all guideline-required sentencing information prior to sentencing.
§ 305.4. Sentence Risk Assessment Instrument procedures.
(a) For each judicial proceeding, the procedure using the SGS Web-based Sentence Risk Assessment Instrument shall be as follows:
(1) Prepare all guideline sentence forms prior to sentencing using SGS Web as required by § 303.1(e):
i. create a Judicial Proceeding;
ii. complete the Prior Record Score module;
iii. complete the Offense module, including all offenses for which the offender has been convicted and are pending before the court for sentencing at the same time;
iv. upon completion of the Offense module, guideline sentence forms are available;
v. for cases identified by the Sentence Risk Assessment Instrument, ''RNR Report Recommended'' is included on the guideline sentence form;
vi. an RNR Report, if prepared, shall be presented to the court for consideration prior to sentencing.
§ 305.5. Sentence Risk Assessment Instrument recommendation—general.
(a) The risk score for the offender and the associated risk category is calculated based on the most serious conviction offense in the judicial proceeding.
(b) Typical risk offenders. For offenders who are identified as typical risk, the Commission makes no additional recommendation.
(c) Low risk or high risk offenders. For offenders who are identified as low risk or high risk, the guideline sentence form will include a notation ''RNR Report Recommended.'' No information on risk category will be provided. Courts are encouraged to adopt procedures to permit the preparation of RNR Reports based on the guideline sentence form recommendation.
(d) Additional information obtained through an RNR Report may assist the court in determining an appropriate and individualized sentence, including the suitability of various sentencing alternatives and programs as well as the duration and intensity of supervision.
(e) An example of sentence guideline form with ''RNR Report Recommended'' is found at § 305.9.
§ 305.6. Offense Types. Classification of Prior Convictions and Current Convictions, including Inchoates.
§ 305.7. Recidivism Risk Scales.
§ 305.8. Recidivism Rates by Risk Score.
§ 305.9. Guideline Sentence Form.
[Pa.B. Doc. No. 19-1094. Filed for public inspection July 19, 2019, 9:00 a.m.]
No part of the information on this site may be reproduced for profit or sold for profit.This material has been drawn directly from the official Pennsylvania Bulletin full text database. Due to the limitations of HTML or differences in display capabilities of different browsers, this version may differ slightly from the official printed version.
