Difference

YLS/CMI vs SAVRY: Youth Risk Assessment Tools Compared

A head-to-head comparison of YLS/CMI and SAVRY for juvenile justice. We analyze predictive accuracy for general recidivism versus violence risk, the role of protective factors, and how each tool guides intervention matching for adolescents.

Get in touch Learn more

Risk analyst performing AI risk assessment on laptop, risk matrices visible, casual office risk session.

THE ANALYSIS

Introduction

A head-to-head comparison of the two dominant youth justice risk tools, framing the choice between general recidivism prediction and specialized violence risk assessment.

The Youth Level of Service/Case Management Inventory (YLS/CMI) excels as a comprehensive case management platform by assessing general recidivism risk across eight criminogenic need domains. Its strength lies in its integration with the Risk-Need-Responsivity (RNR) model, directly linking assessment scores to intervention targets. For example, a meta-analysis by Olver, Stockdale, & Wormith (2014) found the YLS/CMI's total score significantly predicts general recidivism (mean r = 0.25), making it a powerful tool for allocating resources across a broad youth justice population.

The Structured Assessment of Violence Risk in Youth (SAVRY) takes a different approach by focusing specifically on violence risk through a Structured Professional Judgment (SPJ) framework. This results in a critical trade-off: it sacrifices breadth for depth in violence prediction. The SAVRY uniquely incorporates dynamic protective factors alongside historical and social/contextual risk items. Research indicates the SAVRY Summary Risk Rating has a large effect size in predicting violent reoffending (AUC = 0.74–0.78), outperforming tools that are not specifically designed for violence.

The key trade-off: If your priority is a unified platform for general recidivism reduction and intervention matching across all criminogenic needs, choose the YLS/CMI. If you prioritize the nuanced, violence-specific prediction that integrates protective factors and clinical oversight for high-risk adolescents, choose the SAVRY. The decision hinges on whether your system values broad case management efficiency or specialized violence risk formulation.

HEAD-TO-HEAD COMPARISON

Feature Comparison: YLS/CMI vs SAVRY

Direct comparison of key metrics and features for youth justice risk assessment tools.

Metric	YLS/CMI	SAVRY
Primary Prediction Target	General Recidivism	Violence Risk
Protective Factors Assessed
Age Range Validation	12-18	12-18
Total Items Scored	42	30
Dynamic Risk Factors
AUC for Any Reoffense	0.66-0.75	0.67-0.78
AUC for Violent Reoffense	0.65-0.71	0.72-0.80
Structured Professional Judgment

YLS/CMI vs SAVRY

TL;DR Summary

A head-to-head comparison of the two dominant youth justice risk tools. YLS/CMI focuses on general recidivism and criminogenic needs, while SAVRY targets violence risk with protective factors.

Choose YLS/CMI for Case Management & General Recidivism

Best for comprehensive intervention planning. The YLS/CMI is a fourth-generation tool rooted in the Risk-Need-Responsivity (RNR) model. It assesses 42 items across 8 domains (e.g., education, peers, substance abuse) to identify criminogenic needs.

Strength: Directly links assessment to case management by identifying dynamic risk factors that can be targeted for intervention.
Use Case: Ideal for probation officers needing a structured plan to reduce general reoffending, not just violence.
Trade-off: Less focused on protective factors, which are critical for resilience-based youth justice approaches.

Choose SAVRY for Violence Risk & Protective Factors

Best for structured professional judgment on violence. The SAVRY uses a Structured Professional Judgment (SPJ) model with 24 risk items and 6 protective factors, allowing clinicians to integrate case-specific context.

Strength: Explicitly incorporates protective factors (e.g., strong social support, positive school engagement) that can mitigate risk, aligning with developmental maturity principles.
Use Case: Essential for forensic settings, custody decisions, or discharge planning where the specific outcome is violent reoffending.
Trade-off: Does not provide a standardized case management plan for non-violent criminogenic needs.

YLS/CMI: Predictive Accuracy & Implementation

Strong AUCs for general recidivism. Meta-analyses show YLS/CMI total scores predict general reoffending with moderate-to-high accuracy (AUC ~0.65-0.75).

Standardization: High inter-rater reliability due to explicit scoring criteria, reducing subjective bias in scoring.
Limitation: Risk of 'net-widening' if over-reliance on the tool leads to unnecessary interventions for low-risk youth. Requires rigorous training to avoid score inflation.

SAVRY: Predictive Accuracy & Clinical Utility

Superior for violence-specific prediction. The SAVRY Summary Risk Rating consistently outperforms actuarial tools in predicting violent outcomes (AUC ~0.70-0.80) by combining empirical factors with professional override.

Flexibility: The SPJ model allows clinicians to adjust final risk levels based on unique case factors, which is highly valued in court testimony.
Limitation: Professional override can introduce bias if not anchored to the empirical items; requires highly trained forensic clinicians for valid administration.

CHOOSE YOUR PRIORITY

When to Choose YLS/CMI vs SAVRY

YLS/CMI for Violence Prediction

Strengths: The YLS/CMI contains items correlated with violence, but it is not its primary design focus. It may over-classify youth as high-risk for violence based on general antisocial attitudes rather than specific aggressive scripts.

Verdict: Adequate as a general screener, but lacks the specificity required for forensic violence risk assessment.

SAVRY for Violence Prediction

Strengths: SAVRY is purpose-built for this. It structures the evaluation around historical violence, social/contextual triggers, and clinical items specific to aggression. Crucially, it integrates Protective Factors that can buffer against violent outcomes, a feature absent in YLS/CMI.

Verdict: The definitive choice for predicting violent re-offending. Its Structured Professional Judgment (SPJ) model allows clinicians to weigh risk and protective factors to formulate a final summary risk rating, essential for high-stakes court decisions.

Enabling Efficiency, Speed & Accuracy

Intelligent Analysis, Decision & Execution

We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.

Talk to Us

Search across company data

Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.

Useful when people spend too long searching or get different answers from different systems.

Enterprise searchRAGPermissions

Automate internal workflows

Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.

Useful when repetitive work moves across multiple tools and teams.

AI agentsWorkflow automationGovernance

Add AI to products and internal tools

Build assistants, guided actions, or decision support into the software your team or customers already use.

Useful when AI needs to be part of the product, not a separate tool.

AI integrationDecision supportModel routing

METHODOLOGY COMPARISON

Technical Deep Dive: Predictive Validity and Methodological Rigor

A rigorous examination of the psychometric properties, validation studies, and methodological foundations that distinguish the YLS/CMI's risk-need-responsivity model from the SAVRY's structured professional judgment approach for adolescent populations.

Both demonstrate strong predictive validity, but for different outcomes. The YLS/CMI shows superior prediction for general recidivism (AUC 0.66-0.75), while the SAVRY excels specifically for violent reoffending (AUC 0.67-0.80). A 2021 meta-analysis by Vincent et al. found the SAVRY Summary Risk Rating outperformed YLS/CMI total scores for institutional violence prediction. However, the YLS/CMI provides better discrimination for non-violent technical violations and treatment readiness. The choice depends on whether your system prioritizes broad case management or targeted violence prevention.

THE ANALYSIS

The Verdict

A data-driven comparison of YLS/CMI and SAVRY to guide the selection of the appropriate risk assessment tool for youth justice systems.

The YLS/CMI excels as a comprehensive case management platform because it operationalizes the Risk-Need-Responsivity (RNR) model. Its strength lies in identifying a broad range of criminogenic needs—such as education, peer relations, and substance abuse—that are directly linked to general recidivism. For example, a meta-analysis by Olver, Stockdale, & Wormith (2014) found the YLS/CMI total score significantly predicts general reoffending (mean weighted r = 0.32), providing a strong empirical basis for allocating intervention resources across multiple life domains.

The SAVRY takes a fundamentally different approach by focusing specifically on violence risk and incorporating protective factors. This Structured Professional Judgment (SPJ) tool doesn't just sum risk items; it requires a clinician to weigh dynamic risk factors against strengths like strong social support or positive school engagement. Research indicates the SAVRY Summary Risk Rating is a robust predictor of violent reoffending (AUCs ranging from 0.72 to 0.78), often outperforming tools designed for general recidivism when the outcome of interest is specifically a violent charge.

The key trade-off: If your priority is a structured case plan to reduce overall reoffending through targeted interventions, choose the YLS/CMI. If your mandate is to prevent future violence by understanding what protects a youth as much as what endangers them, choose the SAVRY. The decision hinges on whether the system's goal is broad rehabilitation or targeted public safety risk management.

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.

Limited slotsGet a Free AI Consultation

How We Work

Custom AI workflows for your Business

One-fit-all AI don't work for modern businesses. At Inferensys, we aim to understand your business & custom requirements; which we use to define most efficient agentic workflows, the data, and the tools for your business.