| Measure | Ontario corrections ministry | Durham Regional Police police | Thunder Bay Police police |
|---|---|---|---|
| Respondent type | Provincial ministry (Crown in right of Ontario) | Regional police service + board | Municipal police service + board |
| Window | 1984–2026 | 2009–2026 | 2009–2026 |
| Decisions | 484 | 60 | 57 |
| Matters | 344 | 40 | 48 |
| Decisions per matter | 1.41 | 1.37 | 1.19 |
| Counsel signal | 45% | 42% | 70% |
| Self-represented signal | 34% | 25% | 51% |
| Multiple respondents named | 38% | 60% | 49% |
| Median filing year → last decision | 1.0 y | 1.0 y | 3.0 y |
| Longest matter | 9 y | 10 y | 7 y |
| Distinct adjudicators | 74 | 29 | 23 |
| Largest single-adjudicator share | ~6% | 10% | 20% |
| Gate : substance citation ratio | 4.6 : 1 | 5.7 : 1 | 2.8 : 1 |
| Payment orders | 6 | 1 ◆ | 0 |
| Abandoned | 28.8% | 13.2% | 2.1% |
| No final outcome in record | 20.6% | 5.3% | 20.8% |
| Deferred | 11.0% | — | 4.2% |
| Jurisdiction / delay | 3.5% | 18.4% | 14.6% |
| Most-pleaded grounds | Disability 40% · reprisal 23% · race 21% | Race and reprisal tied, 11 matters each | Disability, then reprisal in 26 of 50 applications |
◆ Durham's single payment order does not appear in the tribunal record — the Tribunal cancelled its own award in 2019 and the $10,000 stands by order of the Divisional Court, upheld at the Court of Appeal with leave refused at the Supreme Court. Thunder Bay is counted in application files (50) as well as matters (48), because consolidated decisions break the one-file-one-matter convention there.
The finding that survives all three
70% counsel signal
42% counsel signal
45% counsel signal
Bars show the abandonment rate — matters ending because someone stopped answering.
Across a provincial ministry, a regional police service and a municipal one — institutions with nothing in common beyond a tribunal — the share of matters dying at a paper gate moves inversely with representation. Represented matters get dismissed on the merits of a preliminary test. Unrepresented matters stop answering.
Stated carefully: correlation across three institutions, not causation, and the counsel signal is a text-presence proxy rather than a filed record of representation. A pattern, not a mechanism.
What replicates, and what does not
| Finding | Status |
|---|---|
| Reprisal outranks every substantive ground except disability | Replicates — Thunder Bay and Ontario corrections, two unrelated institutions |
| Gate authorities cited far more than substantive ones | Replicates at all three — 4.6:1, 5.7:1, 2.8:1 |
| Representation predicts the exit route | Replicates at all three, in the same direction |
| Two dockets — insider and public — dying differently | Replicates at Thunder Bay and corrections |
| "The two clocks" — answers slowing while endings accelerate | Thunder Bay only. Corrections shows a 1.0-year median with a long tail instead. Not generalisable until a fourth institution tests it. |
Method
All figures produced by the same parser, the same disposition classifier and the same matter resolution by file number, with zero segment bleed verified. Matters are the unit of analysis throughout; decisions are counted separately and always labelled. The Tribunal does not publish filing dates, so all durations use the year encoded in the file number and month-level precision is never implied.
This table extends by one column per completed case study. Ontario corrections was added August 2026.