1. Scope and responsibility
This policy applies to papers, reports, cases, surveys, publications, course research, public presentations and AI-assisted material published or funded in GFSA's name. The research team manages editorial work; ethics, complaints and conflicts require appropriately independent handling.
2. Content and evidence levels
Government statistics, public datasets, primary sources, repeatable calculations or formally published research. State the source, date, scope and limitations.
State the sample, collection method, period, missing data and likely bias. Do not generalize it to an entire market or universal cause.
May describe process, reasoning and follow-up, but cannot prove general effectiveness.
Identify the source, school and modern use. Do not present it as a mechanism proven by modern science.
Clearly label it as an author's view, discussion or hypothesis, not as “research shows.”
3. Research and scientific language
- Terms such as empirical, scientific, clinical, psychological, physiological, energy, frequency, quantum, big data and predictive model require methods and evidence that actually support them.
- Correlation, subjective feedback and one case must not be written as causation.
- Market figures must distinguish official statistics, commercial forecasts, media estimates and GFSA calculations. Incomparable figures must not be added together.
- Titles, summaries and promotions must preserve material limitations stated in the full text.
4. Authors, review and conflicts
- Identify authors, editors, sources, funders, partner organizations and material interests.
- GFSA publication, editorial review, external expert review and formal peer review require different labels.
- A reviewer must not review their own work, a direct commercial partner or content affected by a material conflict.
- Certificates, membership, sponsorship or advertising are not evidence for a research conclusion or ranking.
5. Cases, participants and sensitive data
- Before collection, explain purpose, publication scope, retention and withdrawal.
- De-identification must address location, occupation, family structure, dates, images and event combinations—not names alone.
- Minors, crisis cases, health, financial, legal, domestic-violence and other high-risk data require enhanced review.
- Do not fabricate, merge or selectively omit material facts to make a case more persuasive.
6. Artificial intelligence and automation
- AI may assist translation, organization, formatting and drafting, but does not replace the author, editor or accountable person.
- Do not enter member, client or case data into an external AI service without authorization.
- AI-generated analysis, charts, citations and conclusions must be checked against primary material.
- Disclose material AI generation or analysis, including its purpose or scope.
7. Corrections, withdrawal and versions
Material error, undisclosed conflict, data misuse, fabricated citation or harmful content may require a correction notice, suspension or withdrawal. A correction must retain its date, reason and principal changes; significant published conclusions must not be silently rewritten.
8. Concerns and appeals
Anyone may raise a concern about research integrity, data use, inaccurate citation, conflict or misleading scientific language. Research concerns follow the process in the Ethics and Consumer Protection Code, including an opportunity for response and appeal.