§ Loading
§ Loading
How we decide what we’ll publish, what we refuse to claim, and what we grade as strong, moderate, or emerging. Every claim on this site is traceable back to this page.
Every claim on this site cites — by author and year, in the body and in the chip, never as “studies show.” The bibliography lives where the claim lives, not in a footnote no one clicks. (Edmondson, 1999) (Hull et al., 2019)
This is what our “Cite or refuse” principle means in practice: we will not publish what we cannot source, and we will name what we have chosen not to publish — so the reader can tell the difference between caution and absence.
We read carefully, grade transparently, and refuse publicly. None of those three lives inside a methodology link no one clicks. This page is that link.
Two Canadian national standards for workforce health, plus the international consensus framework for grading the evidence behind a claim. We don’t invent new measurement frameworks — we apply the standards that already exist.
Canada's national standard for psychological health and safety in the workplace, structured around thirteen psychosocial factors — among them organisational culture, civility and respect, psychological support, clear leadership and expectations, workload management, and recognition and reward. The standard treats psychosocial conditions as measurable workplace properties, not as climate sentiment.
Canada's national standard for organisational disability management. Z1011:20 sets the system-level requirements: governance, intake, communication, stay-at-work and return-to-work coordination, accommodation processes, dispute resolution, and outcome measurement. It treats the workplace itself — not the worker — as the unit of analysis.
The international consensus framework for grading the certainty of evidence behind a claim. GRADE rates each body of evidence as high, moderate, low, or very low quality based on study design, risk of bias, consistency, directness, precision, and reporting transparency.
Team-level psychological-safety claims — “this team is psychologically safe,” “this department’s climate is improving” — require statistical evidence that the individual responses actually represent a shared team property, not just a noisy aggregate of personal opinions. The standard tool is the intraclass correlation coefficient: ICC(1, k). We report it on every publication that draws on team-level survey data, and refuse to aggregate when the value falls below conventional thresholds.
Practically: a team with ICC(1, k) ≥ 0.70 has enough between-team variance to defend a team-level claim. Below 0.70, the team’s mean score is reporting noise — individuals disagreeing with each other rather than a shared team property — and we will not aggregate. We report the value either way.
The capstone instrument set we are building extends this discipline to climate conditions specifically tied to disclosure, accommodation, and brain health. MAWS publishes against an existing peer-reviewed scale (Hull et al., 2019); the three forthcoming instruments are from our Safe to Disclose capstone — currently at validation stage, with first publications targeting 2026.
| MAWS | Masking at Work Scale | LIVE · HULL ET AL., 2019 |
| NDACS | Neurodivergent Disclosure and Accommodation Climate Scale | FORTHCOMING · SAFE TO DISCLOSE · 2026 |
| DBS | Disclosure and Belonging Scale | FORTHCOMING · SAFE TO DISCLOSE · 2026 |
| WAES | Workplace Acceptance of Experience Scale | FORTHCOMING · SAFE TO DISCLOSE · 2026 |
“If individual responses don’t agree enough to call the team a unit, we don’t report a team score. The disagreement IS the finding — and we report it as such.”
Every published finding on this site carries a GRADE dot. The dot reports the certainty of the underlying evidence — not the size of the effect or the strength of the claim.
| GRADE | WHAT IT REQUIRES | WHAT A PRACTITIONER CAN DO WITH IT |
|---|---|---|
STRONG | Multiple converging high-quality studies. Consensus among recent meta-analyses. Effects replicated across populations and contexts. | Cite in adjudication. Use as the evidentiary basis for policy. Defend in peer review. |
MODERATE | A single high-quality study, or a convergent body of moderate-quality studies. Effect direction is consistent; replication is pending. | Cite with caveat. Hedge in print. Use as a basis for practitioner judgment, not for policy without further evidence. |
EMERGING | Theoretically grounded. Preliminary empirical support. Effect direction is plausible but not yet replicated at scale. | Report as a watchlist finding. Treat as a hypothesis to test, not a conclusion to apply. |
A strong dot means we would defend the claim under peer review. A moderate dot means we hedge in print. An emerging dot means we are flagging a hypothesis worth watching, not a finding worth acting on.
Naming the refusals belongs on the methodology page. A reader should know what we won’t claim before reading what we do.
Aggregate team-level psychosocial claims without ICC validation at the survey level.
Call moderate evidence strong because a finding fits our thesis.
Cite a paper to support a claim we wrote before reading it.
Use AI to generate prose without a peer-reviewed source for every factual sentence.
Use the standard MBA change-management canon as evidence — those are consultancy frameworks, not evidence-graded findings.
Treat individual disclosure as a measure of organisational safety — the disclosure rate IS the climate signal, and we report both.