Research Hub

What We Analyzed

Sources: WSIAT Open Data Portal, WSIB Employer Performance Data, Ontario public datasets. 100% open source. No paywalls. No corporate databases. Just facts.

Key Findings: What The Data Shows

What This Means for You

Start Here: Quick Tools (Pick Your Path)

Explore the Data: 5 Interactive Charts

These visualizations let you filter, zoom, and discover patterns in 230,392 tribunal records. Click any chart to explore.

Data Quality: ✅ WSIAT data complete (98,992 decisions). ⚠️ ONSBT estimates (limited public data). 📊 Success rates calculated from real outcomes, not samples. See full methodology →

WSIAT Decision Explorer (1987-2026)

98,992 Ontario workers' compensation appeal decisions now available in structured format. The largest open-source WSIAT dataset in Canadian history.

Dataset Overview

Open Data Access:

• WSIAT Metadata (JSON) - Complete statistics for 98,992 decisions
• Decisions by Year (41 JSON files) - Organized 1987-2026
• WSIAT Data Documentation - Schema, numbering format, comparison table
• WSIAT vs BC WCAT Comparison - 13.4:1 decision ratio analysis
• Deep-Dive Analysis Report - 9 advanced pattern categories
• Keyword Network Visualization - Interactive graph showing issue relationships

Official Data Source: WSIAT Open Data Portal - CSV export parsed and organized for open research.

WSIAT Pattern Analysis: 40 Years of Insights (1987-2026)

Deep-dive analysis of 98,992 WSIAT decisions reveals patterns in legal issues, workload trends, and representative participation across four decades.

Top Legal Issues (Most Common Keywords)

Peak Decision Years (Top 5)

• Year 2000: 4,502 decisions (busiest year ever)
• Year 2017: 4,248 decisions
• Year 2018: 3,969 decisions
• Year 2001: 3,844 decisions
• Year 2016: 3,633 decisions

Most Prolific Vice-Chairs (Top 5)

• R. Nairn: 3,860 decisions
• M. Keil: 3,605 decisions
• J. Moore: 2,981 decisions
• V. Marafioti: 2,484 decisions
• S. Ryan: 2,400 decisions

Key Insight: 3,260 unique vice-chairs identified across 40 years. 100% of decisions include vice-chair metadata, enabling workload analysis and consistency tracking.

Full Analysis Report:

• Complete Pattern Analysis Report - All findings, charts, recommendations
• Analysis Data (JSON) - Machine-readable results

Tribunal Evidence Center (April 2026)

We now publish tribunal findings using a strict evidence model: Tier A (confirmed), Tier B (probable), and Tier C (unresolved), with audit confidence intervals.

Four Ontario Tribunals Analyzed (2020-2026)

Total: 134,920 decisions analyzed (98,992 WSIAT + 35,928 other tribunals). All tribunals use the same tiered evidence framework for transparent outcome reporting.

Open Data Access:

• Strict Evidence Table (JSON) - All four tribunals, A/B/C breakdown
• Tribunal Audit Error-Rate Estimates (95% CI) - Confidence intervals for each tribunal
• Issue Slices Summary - Chronic pain, pre-existing conditions, entitlement denial cross-tribunal analysis
• Connecting the Dots CanLII Keyword Visualization Network - Interactive keyword relationship mapping

Research standard: Tier B is always labeled inferred, and unresolved volume is always disclosed.

🎯 Key Takeaway: If you've been denied benefits or accommodations and you're considering an appeal, the overall data suggests you have a strong chance of success—but it varies significantly by tribunal.

• WSIAT (Ontario Workers' Comp): 100% success rate in our predictive model—but this may reflect data limitations, not actual tribunal decisions. Official WSIAT stats show 65-73% worker success rates.
• BCWCAT (BC Workers' Comp): 86.4% win rate—strong odds if you're prepared with medical evidence.
• Other Tribunals: 84.1% win rate across mixed jurisdictions—consistently high success rates.
• HRTO (Human Rights): High abandonment rate (14% of all cases) suggests procedural challenges—but if you persist, success is possible.

⚠️ Important: These predictions are based on AI analysis of decision text, not official tribunal outcomes. Treat them as indicative patterns, not guarantees. Individual case outcomes depend on evidence quality, legal representation, and specific circumstances.

🔧 API Limitations: Many outcomes remain "Unknown" due to CanLII API restrictions (not a CanLII issue—intentional access limits). We tried: API calls (no outcome field), keyword extraction (non-standard phrasing), web scraping (CAPTCHA + rate limiting), and bulk requests (throttled/capped). To get 100% accurate outcomes, we'd need to manually read each of 137,252 cases individually. Our NLP model predicts these unknown outcomes with 79% accuracy based on case keywords and patterns.

Training Data

• 256,734 labeled examples from 105 tribunal decision files
• 14 outcome categories: Granted, Allowed, Dismissed, Denied, Abandoned, Reconsideration, Allowed - Violation Found, Dismissed - No Violation, Costs Decision, Interim Decision, Settled, Withdrawn, No Jurisdiction, Deferred
• Natural Language Processing: Naive Bayes classifier trained on decision keywords, issue descriptions, and tribunal metadata
• Test accuracy: 79.0% on 3,756 held-out examples (industry-standard train/test split)

Confidence Levels

• High confidence (≥80%): 72.1% of predictions (25,213 decisions) - deployed in search results
• Medium confidence (60-79%): Shown with warning label
• Low confidence (<60%): Not deployed, flagged for manual review

Data Sources

• Outcome Summary (JSON) - Overall statistics
• Outcome by Tribunal (JSON) - Tribunal breakdowns
• Outcome by Year (JSON) - Temporal trends 2020-2026
• All raw decision files: Available in /data/tribunal-decisions/ directory

Open Source Commitment: All outcome prediction data is publicly available. We publish our methodology, confidence scores, and accuracy metrics so you can evaluate the reliability yourself.

When you search for tribunal decisions in the 3mpwrApp, you'll now see outcome badges on every case:

Filter by outcome: Search for "chronic pain" + "Allowed" to find winning precedents. Compare similar cases: See how your situation matches cases that succeeded.

WSIB Claim Guides: What Actually Works

Each guide analyzes hundreds of tribunal decisions to show you exactly what evidence wins claims for your specific injury type. No generic advice—these are patterns from real cases.

Based on: 11,430 total analyzed cases (2020-2026 WSIAT decisions). All guides live now: 19 comprehensive injury-specific guides + 5 legal strategy guides available above.

Ready-to-Use Appeal Templates

Professional appeal letters you can customize in 30 minutes. Each template includes:

• Legal arguments from winning cases (exact language that worked)
• Medical evidence checklist (what documents to attach)
• Employer evidence pushback (how to counter their claims)
• Timeline walkthrough (step-by-step what happens next)

What 11,430 Tribunal Decisions Reveal About WSIB Outcomes

We analyzed every tribunal decision from 2020-2026 using rigorous statistical methods (anomaly detection, co-occurrence analysis, timing analysis). Result: Eight measurable patterns showing systematic process variations across the workers' compensation appeals system.

Understanding WSIB Keyword Patterns in Denial Decisions

WSIB denial letters use technical legal phrases: "pre-existing degenerative condition," "no greater severity than normal," "psychotraumatic disability not established." We extracted every keyword from 11,430 cases to identify patterns in this language.

When Justice Becomes Invisible: The Dark Funnel from 2M Injuries to 11,430 Tribunal Decisions

PROVABLE FACTS: 11,430 tribunal decisions represent only 1,905 decisions/year (2020-2026 average). 91.8% lack outcome metadata in CanLII (10,491 cases = no win/loss data). EXTRAPOLATED FROM PEER-REVIEWED RESEARCH: Institute for Work & Health (15-50% injuries never reported) + Public Health Ontario (1 in 20 workers injured annually) + Ontario workforce (7.5M) = estimated 1.14-2.29 MILLION workers never reaching tribunal.

🏭 ESTIMATED 100,000-200,000 WORKPLACE INJURIES/YEAR IN ONTARIO
         ↓ (50% never reported - IWH research)
  
  50,000-100,000 reported to employers
         ↓ (30% not claimed to WSIB - employer pressure, fear)
  
  35,000-70,000 WSIB claims filed annually
         ↓ (60% accepted but inadequate OR denied)
  
  14,000-28,000 initial denials/disputes annually
         ↓ (75% don't appeal - exhaustion, poverty, despair)
  
  3,500-7,000 internal appeals (reconsideration)
         ↓ (67% resolved/abandoned before tribunal)
  
  📊 1,900 WSIAT DECISIONS ANNUALLY (what we analyzed)
         ↓ (91.8% of outcomes uncategorized)
  
  ⚖️ 156 CASES WITH CLEAR OUTCOMES (what workers learn from)
      

Ontario Human Rights Tribunal (OHRT) Pattern Analysis

Analyzing disability discrimination cases, settlement patterns, and systemic barriers. Expanding beyond workers' compensation to cover employment discrimination, housing, and services.

Estimated Launch: Summer 2026 | Expected Dataset: 5,000+ decisions

Employment Standards Decisions & Wage Theft Patterns

Tracking unpaid wages, termination disputes, and employer violations. Cross-reference with WSIB claims to identify employers systematically denying rights.

Estimated Launch: Fall 2026 | Expected Dataset: 8,000+ decisions

Multi-Tribunal Pattern Detector

Compare outcomes across WSIB, Human Rights, Employment Standards, and Landlord-Tenant tribunals. Identify workers caught in multiple systems, systematic employer bad actors, and regional disparities.

Estimated Launch: 2027 | Requires: All Ontario tribunals collected

Provincial Comparison: BC, AB, QC, NS, MB, SK

Expand WSIB visualization to cover WorkSafeBC, WCB Alberta, CNESST (Quebec), and all provincial systems. Compare denial rates, appeal success, and systemic patterns across Canada.

Estimated Launch: 2027-2028 | Expected Dataset: 50,000+ decisions