Revision and error policy

When a set of statistics is revised, an already published value or a figure is changed. This may happen e.g. because new and more reliable source data has become available, because of changes in methods, classifications and definitions or because of error corrections. Revisions are usually planned, and the users are notified in advance.

Published statistics and other material must be quality assured when published, but errors may occur in figures or related text. When an error is detected, the users are informed, and the error is corrected as soon as possible.

Revisions

Revisions are made in accordance with the ESS guidelines on revision policy for PEEIs (Principal European Economic Indicators). This means that revisions comply with harmonised procedures and principles and that revision practices for the individual sets of statistics are put down in writing and disseminated to the users of the statistics.

The users are notified in advance of scheduled revisions. This is done in the documentation of statistics and sometimes on Statistics Denmark’s website, e.g. in the scheduled releases.

Most revisions occur on a regular basis. This happens when sources required for a set of statistics are not fully available to meet the need for flash estimates. In such cases, preliminary statistics may be prepared on the basis of part of the sources to ensure timeliness. When the planned sources are available, the final figures are prepared. Macro-economic statistics follow the Harmonised European Revision Policy for Macroeconomic Statistics (HERP), which indicates at what intervals and how soon after the end of the reference period revisions can be made. Revisions thus take place in the statistical systems of all countries.

When a set of statistics is seasonally adjusted, it will automatically result in revision of previously released seasonally-adjusted figures when a new period is added. This also happens, even though there are no changes in the historical non-seasonally-adjusted figures. 

Some statistics must be consistent with statistics produced outside Statistics Den-mark. In some cases, there may be interim revisions that do not have an end date, and thus do not exist in an actual final form. This is the case with statistics such as the Emission Accounts. 

Major revisions are especially prevalent in macro-economic statistics that follow binding international guidelines (so-called manuals) and are based on a number of different primary statistics; which is the case for national accounts, government finances, balance of payments and working time accounts. An important quality of these sets of statistics is that they are intercorrelated and they describe the economic trend over many years. Since efforts are also made to ensure consistency between primary statistics and macro-statistics, there may be cases where new sources and methods are only partly incorporated in the primary statistics, or where the most user-friendly solution is for any level changes to await a major revision to ensure coherence between primary statistics and macro-statistics. 

Another reason for making a major revision is the fact that new international guide-lines and assessment rules have been introduced, which imply major changes. To en-sure coherence between the various statistical systems, major revisions in the national accounts, government finances, balance of payments and working time accounts are carried out in accordance with a common schedule, of which users and stakeholders are notified as soon as possible. 

The major revisions are carried out years apart and also change what is referred to above as final figures. For macro-economic statistics e.g., major revisions must be carried out every five years according HERP. 

Statistics Denmark is continuously improving the statistical production to give the best possible account of the situation in society. For that reason, changes may occur in classifications, data sources and methods, where a revision of the statistics is deemed necessary, even though it is beyond both the regular revisions and major revisions. However, the users will always be notified of these changes.

In connection with revisions, it must be ensured that:

• Period and scope of the basis for revisions are determined and coordinated with the remaining statistical production.
• It appears from the documentation of statistics that regular revisions take place, and it is described when the statistics will be released in revised and fi-nal form. It also appears from the documentation of statistics if seasonal adjustments are made.
• For statistics that are subject to regular revisions, it appears from the news series “Nyt fra Danmarks Statistik” (in Danish only) and other publications when preliminary figures are concerned. 
• Analyses are made of the revisions to continuously improve the calculations of the preliminary figures. E.g. the size of revisions is examined and whether they are in a particular direction.
• The scope and any causes of revisions are described in both the documentation of statistics and “Nyt fra Danmarks Statistik”.
• Revisions are discussed with the users on a regular basis.
• In case of an unscheduled revision, changes will be described in the related documentation of statistics, and users will be notified, when this is deemed necessary.
• For some sets of statistics, Statbank Denmark has version tables showing the step by step changes in released figures.

Errors

Errors in the released statistics may be caused by errors in data deliveries from external sources or reporting from enterprises, but they may also be caused by errors in Statistics Denmark’s data processing or dissemination.

In both cases, Statistics Denmark will correct the error as soon as possible. We examine which releases, data, statistics and periods that are affected, what caused the error, and how it can be corrected. Once the error has been corrected, the users whom we know have used the release will be notified as soon as possible. In a few cases, it may be necessary to give notice of the error before it is corrected.

Error corrections in figures or text will be marked and described in the following way:

Nyt fra Danmarks Statistik:

• Corrections are made in the latest version.
• The error will be described, and the correction will be inserted in red.
• A date will indicate when the correction was made.
• Subscribers to the news series will receive an email describing the error.

Statbank Denmark tables:

• Tables in Statbank Denmark can be closed temporarily, if this is deemed nec-essary.
• All periods affected will be corrected as soon as possible – preferably on the same day as the error is found.
• The error will be described in the table footnote.
• An email is sent giving notice of the error to users who have a user profile in Statbank Denmark.

Subject pages:

• When data have been corrected in Statbank Denmark, the relevant subject page will also be updated.

Publications, Statistics Denmark analyses, news articles:

• Only the digital version will be changed (publications).
• Only the latest version will be changed (publications).
• The error will be described, and the correction will be inserted in red.
• A date indicates when the correction was made.
• Subscribers to the news series will receive an email describing the error.

Press releases:

• If there are errors in a press release, an estimate will always be made of the scope and timeliness of the error as well as whether important principal conclusions are affected by it. If it is deemed necessary, a new notification will be distributed from which it appears that a correction is concerned. The corrected version will be published via Ritzau.

Facebook, Twitter and LinkedIn:

• Posts on Facebook, Twitter and LinkedIn will not be deleted, instead the text in the post will be changed (Facebook and LinkedIn).
• From the beginning of the post or thread, it will appear that it has been changed, and the reason for the change will be described (Facebook and LinkedIn).
• If a wide-ranging error is concerned, a new tweet will be published (Twitter).