Accuracy and reliability

Overall accuracy

At the time of the first publication of quarterly figures, reports are typically missing from 5.4 percent (on average) of the units included in the statistics. By the fourth publication (one year later), this figure decreases to 1.5 percent (on average). Follow-up procedures for non-response prioritize the largest enterprises.

Sampling error

Not relevant for these statistics.

Non-sampling error

The statistical frame population consists of legal units with at least 10 full-time employees (cut-off). This covers approximately 93 percent of total turnover in mining and manufacturing. The published data are not grossed up to cover the full target population, and units below the cut-off are therefore excluded. Consequently, the design does not provide coverage of enterprises with fewer than 10 full-time employees. The frame population is delineated using industry and employment information from the ESR. The quality of these sources is generally high, resulting in only minor coverage errors. Each year, 100–200 new units are included, of which 20–30 are subsequently exempted—most often because the unit had been assigned an incorrect industry code and is in fact not a manufacturing enterprise.

Respondents are required to allocate their product sales across very detailed product codes from the Combined Nomenclature, which contains around 10,000 codes. Reporting under incorrect codes is therefore a common measurement error that is not always detected—although checks against industry codes, atypical unit values, user feedback, and other controls identify a share of such errors. This type of error has the greatest impact at the most detailed level; when codes are aggregated to, for example, 6- or 4-digit levels, the effect is reduced, since most incorrect codes are still relatively close to the correct ones in the nomenclature.

In 2013, a study was conducted comparing the data with External Trade in Goods statistics at the enterprise level. This indicated a tendency for enterprises in the Sales of Goods in Manufacturing survey to use fewer product codes than they do when reporting to External Trade in Goods.

When respondents use incorrect product codes, the reasons are either lack of knowledge about classification or the fact that allocating sales across codes is too time-consuming. The latter is particularly relevant for enterprises producing many different and changing products, e.g. subcontractors producing to order. There is typically some degree of overreporting on so-called “residual codes,” such as “articles of iron and steel, not elsewhere specified”—in the sense that sales are reported under such codes that should more appropriately have been placed under a more specific code.

Quality management

Statistics Denmark follows the recommendations on organisation and management of quality given in the Code of Practice for European Statistics (CoP) and the implementation guidelines given in the Quality Assurance Framework of the European Statistical System (QAF). A Working Group on Quality and a central quality assurance function have been established to continuously carry through control of products and processes.

Quality assurance

Statistics Denmark follows the principles in the Code of Practice for European Statistics (CoP) and uses the Quality Assurance Framework of the European Statistical System (QAF) for the implementation of the principles. This involves continuous decentralized and central control of products and processes based on documentation following international standards. The central quality assurance function reports to the Working Group on Quality. Reports include suggestions for improvement that are assessed, decided and subsequently implemented.

Quality assessment

The main non-sampling error is the measurement error concerning classification at the most detailed CN level, as respondents do not always report sales according to the correct codes. Furthermore, data on quantities are generally less reliable than those on values, as some respondents estimate quantities and other do not answer, so that estimations must be made in the statistical production process.

Data revision - policy

Statistics Denmark revises published figures in accordance with the Revision Policy for Statistics Denmark. The common procedures and principles of the Revision Policy are for some statistics supplemented by a specific revision practice.

Data revision practice

With each release of data for a new quarter, data for previous quarters are also released in revised form. Data are always provisional at the first release.

In theory, data for Manufacturers' sales are never final. This means that it is always possible to revise the data if significant errors are found. In practice, data are not revised indefinitely, but are revised in accordance with the following guidelines:

• Late data reports are always incorporated, but normally data are never reported more than one year late
• When Statistics Denmark finds errors in reported data, they are corrected. Errors are not always detected before the first release, especially errors regarding the detailed distribution according to CN codes and the data on quantities. Errors are normally corrected in all quarters of the year when they are found, plus in all quarters of the three previous years.
• Once a year, a revision of industrial classification for reporting units is carried out. Based on reported CN codes and other sources, some units are allocated with a different classification. This is implemented with the first release of the fourth quarter each year, but the three previous quarters are also revised with the altered classification. Normally, between 30 and 70 units change classification.

Data older than the current year plus three previous years are thus only revised in very special cases. The release of 2nd quarter 2014 has been the only recent case of this type, in which turnover for manufacturing of pharmaceuticals was revised for the period 2005Q1-2014Q1, due to changed reporting's that improved the coverage.

Data on manufacturer's sales of goods are submitted to Eurostat annually for the previous reference year. If major revisions have been made since the last data submission, a revised version of previously reported data is also submitted.