Evidence standards
Confidence begins
with clarity.
How we keep research findings, practical interpretation, and unanswered questions distinct.
How to read an Insight
Five labels.
Different kinds of claims.
These labels describe the role of a statement. They are not a ranking or a guarantee that a finding applies everywhere.
- 01Established evidence
- Supported within the boundaries of the cited evidence, not a universal guarantee.
- 02Emerging evidence
- Early, limited, or not yet replicated findings.
- 03Workforce Nexus interpretation
- Our synthesis or practical application, not a result directly tested by the source.
- 04Hypothesis
- An explanation to test, not an established finding.
- 05Proposed framework
- An organizing model without a scientific validation claim.
The editorial standard
What every claim
should be able to withstand.
Traceable evidence, proportionate language, and a clear account of what remains uncertain.
Start with the question
Define the issue before selecting a solution.
Examine possible mechanisms and compare alternatives, including changes that do not involve buying a product. Assess employee and organizational outcomes separately before describing shared value.
Keep sources attached to claims
A citation should support the statement beside it.
Consequential factual claims need traceable references. Preserve the relevant population, geography, denominator, measurement basis, and reference period. The period data describe may differ from their publication date. Source reputation does not replace examining the method.
Match the language to the design
Association, impact, and scenarios are different.
Describe observational findings as associations. Causal language requires an appropriate research design and its limitations. A before-and-after change alone does not establish program impact. Scenario calculations are not forecasts or promises of realized savings.
Review before publication
Make dates and material corrections meaningful.
Articles remain unpublished until source checks and human editorial review are complete. Publication and substantive revision dates must reflect real events; material corrections belong in a visible revision history. We do not display individual bylines. Internal review is not external peer review.
Disclose commercial boundaries
Evidence of a problem does not prove a product solves it.
Distinguish educational content from consulting, insurance, and other regulated activity. Recommendations should identify relevant commercial interests and competence boundaries. Do not present a commercial offering as the necessary conclusion of a research finding.
Protect people in future research
Establish governance before collecting original data.
Employer or worker data collection requires agreed definitions, consent, access controls, retention rules, and disclosure safeguards. A universal minimum reporting-cell size has not been adopted. No proprietary workforce index or validated assessment is claimed.
Method before claims
Research develops.
The boundaries stay visible.
Original studies, longitudinal research, and proprietary assessments remain future directions, subject to appropriate governance, design, and validation.