Code of Practice

A figure is only worth as much as the process behind it. Every published number can be traced to who committed it, when, and from what source.

• No one person can both prepare and publish a figure. An editor uploads and validates; an administrator commits. The system enforces this, not a convention.
• Every commit, revision, deletion, freeze and publication is written to an append-only audit log.
• Changes to published figures appear on the public revision log, whether or not anyone asks.
• Issuing or withdrawing access to the data bank is recorded against the person who did it.

Statistics are published with their limits attached. Where the data bank cannot support a claim, it says so rather than filling the gap.

• Each series carries its source, its unit, its methodology version and its reporting cadence.
• Cadence is read from the records themselves, not from a registry field, so a series is never described as more frequent than it is.
• The database enforces one figure per series, period and region. A second figure for the same period is rejected, not added.
• Where a period holds no record, that is shown as absent rather than as zero.
• The planning model marks every input as measured, derived or assumed, and never presents an assumption as a measurement.

Data is published to be used. Access is open by default and restricted only where the law or the privacy of a respondent requires it.

• Published series are available without an account, through the site and through the public API.
• Every chart and table can be exported.
• Restricted material is restricted for a stated reason, and the reason is shown to the person refused.

Publishing early and revising later is normal practice. What makes it defensible is saying which state a figure is in.

• Figures are marked provisional, revised or final.
• A revision changes the figure and records the old value beside the new one.
• A published bulletin edition is frozen. Corrections are issued as a later edition, never by editing a published one.
• A period can be locked so that no further change is possible without a super administrator reopening it, which is itself recorded.

Data collected from households and businesses is used to produce statistics and for nothing else.

• Assessment responses are published only as aggregates, never as individual records.
• Aggregates are withheld entirely until a minimum number of responses has been collected, so no single respondent can be identified.
• Access to identifiable assessment data is granted per person, for a stated purpose, and every view is logged.
• Personal data is handled under the Nigeria Data Protection Act 2023.

• Independent external review of methodology. Series methodology is published; it has not been externally assessed.
• A formal complaints and appeals route for a disputed figure. Corrections are currently raised through the data request channel.
• Pre-release access arrangements. No figure is currently released to any party ahead of publication, and there is no list of people who receive one.