When to use ‘in development’ for official statistics?
Existing official statistics
The Code endorses innovation and improvement in Principle V4. Producers will need to decide whether changes being made to existing official statistics merit the statistics being flagged as ‘in development’ when published.
The key consideration is the expected impact of the development on the official statistics – do users need to be alerted to the potential effect to ensure that they use the statistics appropriately?
If the changes are relatively minor, then there is no need to use the ‘in development’ label.
New official statistics
For new statistics, producers may first test a new method or the suitability of data sources to see whether the new statistics are appropriate estimates and that the new approach is working as expected.
It is good practice for a producer to publish research findings that describe the new approach, so being transparent and supporting user understanding. This early testing of the method would not be a publication of official statistics in development.
Instead, official statistics in development should be published at the point the producer is content that the method is producing sufficiently robust figures that are an acceptable representation of the subject being measured.
Using and removing the ‘in development’ label
The label ‘official statistics in development’ should be used to alert users to statistics that may be affected by a development, for example, leading to a wide degree of uncertainty in new estimates or increasing the uncertainty in existing statistics. When producers are sufficiently confident that the new approach is working appropriately and as expected, they can then publish the statistics as ‘official statistics’.
Producers should explain the nature of the development – build users’ confidence in the estimates by being open about what has been done to produce them. The development could relate to new statistics or to major developments to existing statistics. In describing the development, producers can outline, for example:
- what the statistics are
- why the statistics are needed
- the data sources and methods used
- any coverage or quality concerns
- information about the nature of testing and its outcome