Guidance

Defra Policy Statement on Revisions and Corrections

Published 5 February 2020

Applies to England

1. Introduction

Defra (the Department for Environment Food and Rural Affairs), with its Arms’ Length Bodies (ALBs), follows the requirements set out in the Code of Practice for Statistics (the Code) when publishing our statistics.

We are committed to maintaining public confidence in the independence and integrity of the statistics we produce. Our focus is to be open and transparent about any revisions and corrections we make to our published statistics and will ensure that customers have easy access to comprehensive information about those changes.

In accordance with Practice T3.9 of the Code, we will publish this statement that sets out how we handle scheduled revisions or unscheduled corrections to statistics and data. It will be adopted by core Defra statistics teams and by any teams in our ALBs who wish to do so in preference to publishing their own policy on revisions and corrections.

2. Scheduled revisions ^{[footnote 1]}.

Routinely scheduled or planned revisions take place when we publish timely provisional estimates before all data sources have been delivered or provide regular updates to published statistics. For example:

• ‘Total Income from Farming’ figures are originally published as ‘provisional’ and then updated later to ‘final’.
• Regular updates to individual indicators within the ‘Statistical Digest of Rural England’ are published at regular intervals throughout the year.

3. Unscheduled corrections ^{[footnote 2]}.

Where we have decided to produce statistics we endeavour to provide the best figures we can in light of the available information and resources. Unscheduled corrections may be necessary for a number of reasons:

• Data changes: This type of revision can arise, for example, where information given by data suppliers missed the original publication deadline or when data suppliers give notification that they have submitted incorrect information.
• Errors: Occasionally, errors will occur in our statistical processes. Precautions are taken to minimise these errors and we continually review our processes and procedures to minimise risk. Errors will be corrected promptly and users informed.
• Methodology changes: This type of revision occurs when major changes to administrative systems or statistical methods are put in place. These usually result in improvements to the quality of the statistics and may be backdated to bring historical estimates up to the same standard.

4. Handling Revisions and Corrections

Our aim when handing revisions and corrections is to be transparent with our customers about the need for the changes, when to expect scheduled revisions and how unscheduled corrections will be communicated.

Decisions on how to address any changes to published statistics will be made by the senior statistician in consultation with the Head of Profession as appropriate, see Practices T2.3 and T5.6 of the Code.

Our overarching commitments in handling revisions and corrections are:

• Promptly identify the affected statistics.
• Where corrections are significant, keep the Head of Profession and press and communications teams informed.
• Provide a clear explanation of the reason for the details of the size and impact on the statistics, see Practice Q3.4 of the Code.
• Publish the revised / corrected statistics with a clear explanation of the reason for the change with details of the size and impact on the statistics as soon as is practicable, see Practices T3.9 and Q3.4 of the Code.
• Include a direct comparison of original and revised estimates.
• Where practicable within timescales seek comments from key customers on proposed changes to methodology and on the options for maintaining a consistent time series, including any revisions of previously held data.
• Consult with key customers on any major changes to survey methodology.
• Keep customers informed.

4.1 For scheduled revisions

All scheduled revisions will be treated as other official statistics and their publication date will be pre-announced on the Release Calendar, see Practice T3.1 of the Code.

For key National Statistics that are subject to a scheduled pattern of revisions, where appropriate, we will provide a specific revisions policy within that output.

4.2 For unscheduled corrections

Where major corrections are required, the publication date will be pre-announced on the Release Calendar.

As soon as possible after we establish that a correction is warranted, we will amend all electronic publications which contain the affected statistics, and remove superseded versions.

If a correction is minor or insignificant we may make the changes without alerting users but we will include a note in the publication about the correction.

If a correction is substantial or significant we will issue a prominent alert on the website to notify users about the change and the reason for its occurrence. If the error is such that it could impact significantly on the use of the statistics, we will endeavour to alert key users prior to making the correction.

4.3 For data changes

A judgement will be made as to whether the change is significant enough to publish a revised statistical release as soon as possible or wait until the next planned release. This will be partly dependent on the frequency of releases for that output.

4.4 For errors

These will be corrected promptly.

4.5 For methodology changes

The decision will be taken whether a revised statistical release should be published immediately or wait until the next planned release.

5. Authority

This policy statement has been drawn up by Defra Group Head of Profession for Statistics, who is also responsible for ensuring the day-to-day implementation of these arrangements in the core Department and providing professional oversight and guidance to our ALBs.

Any questions relating to the application of these arrangements to specific releases should be addressed to the statistician responsible for the issue of the release, as shown on the release itself, in the first instance.

Any enquiries about the contents of this statement please emailStatisticsHoPDefra@defra.gov.uk

23 January 2019

1. Scheduled revisions are defined as ‘Planned amendments to published statistics in order to improve quality by incorporating additional data that were unavailable at the point of original publication’. ↩

2. Unscheduled corrections are defined as ‘Amendment made to published statistics in response to the identification of errors following their initial publication’. ↩