Research Standards
CGS USA aims to publish work that is accurate, transparent, proportionate to the evidence, and clear about uncertainty. The method and level of review vary by format, and each publication should tell readers what process it received.
Our standards
- Research questions and methods should fit the public-interest purpose of the project.
- Sources should be traceable and represented fairly; material counterevidence should not be concealed.
- Quantitative claims should identify data sources, units, time periods, and material limitations.
- Interviews and human-participant research should use appropriate consent, confidentiality, security, and ethics review.
- Authors must disclose relevant financial, professional, political, institutional, and personal interests.
- Funders and partners may contribute expertise or review for factual accuracy, but may not require predetermined findings or suppress unwelcome results.
- Editorial changes should improve accuracy, clarity, fairness, accessibility, and legal compliance without changing an author’s conclusion without agreement.
- Substantive errors are corrected transparently; routine spelling or formatting fixes need not receive a public note.
- Generative AI, translation, or automated tools must not be credited as authors. Human authors and editors remain responsible for accuracy, originality, confidentiality, and disclosure.
What each format receives before publication
Research report
Research lead review, subject-matter review, methods review where relevant, fact check, editorial review, disclosure check, and final approval.
Policy brief
Subject-matter review, source and claim check, editorial review, and disclosure check.
Working paper
Author accountability plus editorial and disclosure review; clearly labeled as work in progress.
Commentary
Author accountability, identity and conflict check, source review for material factual claims, and editorial review.
Event report
Review against recordings or notes, speaker attribution check, and clear description of the event’s record status.
Found a problem in our research?
Substantive errors are corrected transparently and noted on the publication. Tell us what you found and where.