Policy on revisions of DWP Official Statistics
Updated 21 December 2018
Introduction
The Department for Work and Pensions (DWP) complies with the UK Statistics Authority Code of Practice for Statistics.
We’ve published this policy in accordance with T3.9 of the Code of Practice.
It explains our policy on:
- revising statistics
- informing users when we make revisions to statistics
DWP aims to avoid the need for revisions to publications unless they are absolutely necessary. We have systems and processes in place to minimise the number of revisions that need to be made.
A “revision” is defined as any change to a public domain official statistic. A public domain release includes:
- printed paper documents
- releases via the internet, Stat-Xplore or Nomis online tools
- releases via a direct question (for example, a Parliamentary Question or Freedom of Information request)
Within our statistical publications, there are 3 principal reasons why we might need to make revisions or changes to published official statistics. These changes can be divided in to:
- scheduled revisions, which are part of the defined methodology for a particular series
- methodological changes that affect published numbers
- revision of errors in our statistical systems and processes
This policy statement describes how DWP will handle revisions and give confidence that all types of revision will be handled in a transparent manner.
Scheduled revisions
Scheduled revisions are built into the defined methodology of a statistical series. These might occur as part of the normal release practice for a given series:
- where initial or provisional estimates are released which are always subsequently revised
- as a result of regular seasonal adjustment, benchmarking or rebasing exercises
Methodological changes
Sometimes, published series are subject to a “one-off” methodological change. This may happen when, for example, a data source or a methodology is improved and changes to historical numbers are significant enough that a revision of previously released tables is needed.
Revision of errors
DWP do perform extensive quality assurance on raw data and statistical tables prior to publication to minimise the possibility of these errors occurring, but occasionally an error slips through. In the rare case when an error occurs, we thoroughly investigate how it occurred and strengthen our processes where necessary.
Key principles for handling revisions
Our general policy is to be transparent with users about the need for revisions, how and when to expect scheduled revisions and how these will be communicated to users.
For scheduled revisions:
- any official statistics outputs which are subject to scheduled revisions will contain a statement setting out how revisions will be handled
- revisions will be either pre-announced in the statistics first release one publication cycle before the change is expected, or on the GOV.UK DWP statistics page
- for major revisions, users will be able to understand the nature of the scheduled revision in advance and its likely impact on the statistics – the revision will be explained again in the release itself, when published
- in some cases, scheduled revisions will affect numbers that have been provided in Freedom of Information requests or Parliamentary Questions – these will not be revised as the original response will have flagged the later release of revised data
For methodological changes and revision of errors:
- any misleading or erroneous figures will be footnoted or removed
- the nature and reason for the change and the likely impact will be explained
- we will announce the date for the release of revised numbers or our intention to change the methodology in advance
- the revised numbers will be released according to the pre-announced date, in compliance with the same principles as the original information
For all revisions:
- revisions will be assessed for disclosure risk before release, taking into account the possibility that some of the data in the revised output have already been released
- revisions will be assessed to balance the benefits (in terms of quality and accuracy) against the costs (including both costs to government and data suppliers), taking into account the expected uses of the statistics
Should we decide not to implement a particular revision, this will be noted in the next issue of the relevant publication, together with the reason for the decision.
Contact information
Contact DWP by email: [email protected].