Chapter navigation
Before publication: reduce the risk
Use these actions while planning, developing and quality assuring the statistics. The aim is to make misunderstanding less likely before the release is published.
Build trustworthy foundations
Be open about what you did and why:
Explain methods, decisions, limitations, revisions and quality issues so users can understand the evidence behind the statistics.
Communicate transparently:
Provide context, caveats and explanations proactively so headline figures are not interpreted in isolation.
Show how quality has been assured:
Make uncertainty, bias, limitations, peer review and improvement activity visible.
Record decisions and accountabilities:
Keep evidence of methods, quality assurance, user engagement, risk judgements, release practices, revisions and improvement actions.
Support use
Understand who uses the statistics and why:
Engage direct, indirect and potential users, including intermediaries and under-represented groups where appropriate.
Design outputs around public value:
Explain what decisions the statistics support and how users should interpret them.
Test the message:
Work with communications colleagues to test headlines, visuals, caveats, user journeys and audience reach routes before release
Make accurate reuse easier:
Provide clear definitions, accessible outputs, metadata, explainers, briefing material or questions and answers where these would help users and intermediaries reuse the statistics accurately.
Anticipate misuse and respond effectively
Identify likely risks:
Consider whether definitions, comparisons, uncertainty, seasonal effects, headline figures, public debate or known misconceptions could lead to misunderstanding.
Build safeguards into the release:
Add proportionate context, caveats, definitions, uncertainty and links to related measures where needed.
Agree roles before publication:
Decide who owns monitoring, clarification, correction, escalation, external engagement and post-release review.
Plan monitoring proportionately:
For sensitive, high-profile, novel or contested statistics, agree what will be monitored, who will monitor it and what would trigger further action.