Durham: 12 of 37 matters (32%)
Twelve corpus matters plead at least one of race, colour, ancestry, ethnic origin or place of origin. Where those matters sit is the finding:
civilians and policing services
Matters can be flagged in more than one social area, so the rows sum above 12.
Race-grounded complaints against this police service are overwhelmingly about policing the public — 11 of 12 — not about employment. That is the mirror image of Thunder Bay, where the race cluster sits inside a record that is 62% internal. Two services, two entirely different discrimination problems, and any provincial figure that averages them describes neither.
Their endings: two dismissed as abandoned; two for no reasonable prospect; two for no Code nexus; one for incomprehensible pleadings; one dismissed after a merits hearing; one proceeding with no final outcome recorded; one under a filing prohibition — and one upheld: the racial-profiling finding that is the only remedy in either corpus, cancelled by the Tribunal, restored by the Divisional Court, and kept only through four courts. Ten of the twelve applicants (83%) never had any representative.
Thunder Bay: 19 of 50 applications (38%)
Social-area coding for the Thunder Bay corpus is text-derived and matter-level coding is pending; the split below is reported at the level the texts support.
Nineteen applications engage the race–ancestry–ethnic-origin cluster — the corpus where two 2018 oversight findings had already documented anti-Indigenous systemic racism. Not one of the nineteen has reached a merits hearing. The Indigenous-specific record — the chair, the officer, the father, the civilian — is read closely on Four faces, one institution, with the federal comparison on One claim, two systems.
The province's own finding: the OHRC's Toronto inquiry
Our two corpora are small. The Ontario Human Rights Commission's public-interest inquiry into the Toronto Police Service is not — and it is the clearest demonstration in Canada of what the s. 31 and s. 35 powers can produce when the Commission actually uses them. Using its inquiry powers to compel Special Investigations Unit case files, and expert analysis by Dr. Scot Wortley, the Commission found that between 2013 and 2017 Black people made up 8.8% of Toronto's population and:
Two findings from that inquiry bear directly on this site's method. First, the SIU cleared officers in over 90% of investigations regardless of the complainant's race — which the Commission was careful to explain does not answer the discrimination question at all, because the SIU looks for criminal wrongdoing, not discrimination, on a criminal standard, and has no power to make Code findings. An exit is not a verdict, in that system as in ours. Second, the SIU did not collect race data; researchers had to reconstruct it from investigator notes and police documents. The absence of the data was itself the finding.
The Commission's broader consultation, published in 2017, gathered nearly 1,650 responses province-wide and found racial profiling reported far beyond policing — in retail, education, health care, child welfare, transportation and housing — with over half of racialized and Indigenous respondents reporting they had experienced it, against 11% of White respondents.
Source: Ontario Human Rights Commission, A Collective Impact (interim report, 2018), including the Wortley Report; and Under Suspicion (2017). Figures are the Commission's and its expert's, restated in our own words; the coding of race in that dataset was deliberately conservative, so the disparities may be understated.
When the Commission acts, this is what it looks like
The OHRC has litigated race cases against police, not only studied them — Phipps v. Toronto Police Services Board, where a Black letter carrier was found to have been racially profiled while delivering mail, was upheld at the Divisional Court and the Court of Appeal, and remains a foundation of the profiling jurisprudence. That is the same institution, the same statute, and the same powers we ask about on the s. 35 page — which is why the question there is not whether the Commission can act against a police service, but why it has not done so in the cases this site documents.
Reprisal, by social area
The same breakdown for reprisal — punishment for asserting Code rights — at Durham. Eleven matters plead it, and it follows the same shape:
Who alleges what: race is a civilian claim, disability is an employee claim
Breaking both corpora down by who is complaining produces the cleanest pattern in the entire dataset — and it holds across two services with opposite records.
Durham — 12 race-cluster matters, by social area
civilians complaining about policing
the service's own people
Eleven of twelve race-grounded matters at Durham are service-docket claims by civilians. Within Durham's 22 services matters, race is the single most-pleaded ground (10), then reprisal (9) and colour (8). Within its 5 employment matters, the top grounds are reprisal, record of offences, race and disability — two matters each.
Thunder Bay — 50 applications, split by who filed
It also means a headline like "38% of applications allege race" is close to meaningless without the split: at Thunder Bay it is 47% of civilian applications and 19% of employee ones. The board chair's five applications plead every one of race, ancestry, ethnic origin, disability and reprisal — the only filer in the corpus to do so.
Unit: Durham matters (12 race-cluster of 37, multi-flagged by social area); Thunder Bay applications (50, categorised by filer). Text-derived; pleading-level verification pending.
Reading it against the provincial jurisprudence
The law these applicants invoke is settled at the appellate level: racial profiling can be proven by circumstantial evidence and inference (Phipps, Pieters), and a claimant's credibility is not diminished by the assertion of discrimination itself (Maynard). The gap this page documents is not doctrinal — the law exists. It is procedural: in these two corpora, race-grounded claims overwhelmingly end at gates where the law of profiling is never applied to any facts.
Method: grounds from decision texts (Durham: pleading-verified per decision; Thunder Bay: text-engaged per application, pleading-level pass pending). Outcome and representation joins from the relational dataset. This page widens per completed case study; no provincial rate will appear until a provincial full-text corpus exists.