Statistical Data Revision Policy

This document definesa statistical data revision policy of National Statistics Office of Georgia (Geostat). It describes the necessity of revision, reasons and types of revision and its principles and appropriate terminology. The document incorporates common guideline principles and subject matter divisions may use them when defining revision policy for a specific statistical product.

What does Revision Mean

Revisions are broadly defined as any change in a value of a statistics released to the public. They canoccur in below listed cases:

  • New observations (one additional month, quarter or year) become available
  • Some current and/or past values are modified
  • Methodology has changed

Data are generally revised in order to incorporate new, improved information.

Why Revision is Needed

Data revisions are a common phenomenon and their goal is to continuously enhance the quality of statistical data. Quality of statistical information comprises several criteria of which two are worth noting: accuracy and timeliness. Both are of particular importance to ensure that statistical information is relevant, or in other words, statistical information meets users’ needs. The need to carry out revisions reflects the obligation of Geostatto produce high quality statistics that are as up-to-date as possible.

Revision policy is based on international standards and requirements. European Statistics Code of Practice – principles 6, 8 and 12 define the revision policy.

  • Principle 6,Indicator 6.6 -"For bigger revisions and changes in the methodologies information is providing at advance”;
  • Principle 8,indicator 8.6 - "revisions are making according the standards, good established and transparent procedures and principles”;
  • Principle 12,Indicator 12.3 - "Revisions are regularly analyzed for improvements of the statistical processes".

When mistakes are discovered, data are modified or methodology has changed, it is critical that the Geostat notifies the public as soon as possible and provides satisfactory explanations to reassure users that the changes where necessary. Thus enabling them to distinguish mistakes made from cases of misreporting. Revision policy is an important aspect of good governance in statistics.

Possible Causes of Revision

  • Modification of estimated statistical data obtained through estimation – timely production of statistical data requires to make estimations of certain data (which are not available by the dissemination date). Since it is generally the case that the estimated values do not correspond precisely to the late incoming data, the original figures are revised;
  • Benchmarking also gives rise to changes. Benchmarking is defined as the adjustment of (generally) higher frequency data to take account of more complete lower frequency results, which become available only later (for example, to modify quarterly data based on yearly data);
  • Classification changes are another, albeit rarer, trigger for revisions. They cause items to be redefined so that the results for the old and new items are different, if they are even comparable at all;
  • Conceptual and methodological changes (e.g. changes in concepts, definitions, classifications etc.) and Improvement of the methodology applied for the production of statistics causes the revision;
  • Revisions can also be caused by changes to the base year, which together with new weighting systems or new projection boundaries, can also alter the paths of time series;
  • Changes in data sources (e.g., availability of new source data that more closely match the concepts, replacement of survey data with administrative data, data of new census etc);
  • Alterations to data may also stem from changes made by respondents to correct erroneous reports, also correction of data initially transmitted by administrative sources could happen;
  • Correction of mistakes – errors made during the compilation process also necessitate revisions to published statistical results.

Planned and Unplanned Revisions

Planned Revision

Planned or scheduled revisions refer to routine revisions and major ones:

  • Routine revisions are changes in published data which are related to the regulardata production process (e.g. estimated values for missing responses are replacedby reported figures or preliminary data are revised and updated accordingtothe schedule).
  • Major revisions are changes in published data, often substantial, which are due toone of the following reasons:
  • Availability of new structural data that are collected at long intervals (usually 5 to 10 years), such as data coming from censuses, labor cost surveys, etc.;
  • An update of the weights or the base year of an index series, often every five years;
  • A change in methodology, concepts, definitions and/or classifications used to produce the series (adoption of a new classification, changes in international statistical standards, etc.);
  • The changes in current legal and normative actsand/or theenter in force of a new legal act;
  • Disclosure of important errors.

Usually data producers take the opportunity of a forthcoming major revision to introduce methodological improvements. This is considered as good practice as it avoids revisions occurring too often. Therefore it is common that major revisions are not determined by one single cause but by a combination of them.

Unplanned (Non-scheduled) Revisions

Unplanned (Non-scheduled) revisions are not announced in advanceeither because they are a result of unforeseeable events such as errors or accidents.

Revision Policy Principles

  • Geostat carries out its responsibilities in respect of revisions by following the current revision policy based on the European Statistics Code of Practice and the ESS guidelines on revision policy in force.
  • The revision policy is published and maintained on the Geostat website.
  • A schedule for the planned revisions is compiled by Geostat and placed on the Geostat website. In addition, a relevant text on the planned revisions of statistics is included in the Annual Statistical Work Programme of Geostat.
  • The users are promptly informed on significant errors identified in the published statistics, leading to non-scheduled revisions. The revised results are released without any delay in an open and transparent manner. The reasons for carrying out the non-scheduled revisions are also published.
  • Major revisions and non-scheduled data revisions are accompanied at the time of publication by relevant documentation
  • Geostat announces substantial methodological changes in the compilation of statistics before the release of statistics based on the new methods.
  • Geostat carries out revision analysis on a regular basis, by applying internationally recognized best practices and prepare relevant documentation
  • Geostat periodically consults users to elicit views about its revision practices.

Dates of Revision

Preliminary data revisions and major revisions are made according to the revision calendar/schedule

In special cases, when important mistakes are identified resulting in data incompliance, revisions are made immediately.

1