How are codes shown in dataset tables?

Dataset tables show codes as breakdowns: rows or columns headed by a section, division, group or class, with figures for each heading, and metadata stating the classification version, the reference period, and the statistical unit. A table without that metadata cannot be compared with another table, because the reader cannot tell whether the two tables use the same version, describe the same period, or count the same kind of unit.

The code headings in a table are labels for aggregates, not entries for individual businesses, and a figure in a cell is the total for the units in that category, not a statement about any one unit.

What does a dataset table show at each level?

A dataset table can show codes at any level, and the level must be stated in the table heading or the metadata. A section level table shows broad areas, such as manufacturing or construction, with one row per section letter. A division level table shows two digit divisions inside a section. A group level table shows three digit groups. A class level table shows four digit classes, the most detailed level. The Central Statistics Office notes that National Accounts uses particular groupings based on NACE, labelled A10, A21, A64 and A88, when publishing data, which are aggregations, not NACE levels themselves. A table that shows an aggregation should label it as an aggregation, with the groupings defined in the metadata, so that a reader does not mistake an A10 total for a section total, or a division total for a class total. The level of the table decides what can be compared: a section table can be compared with a section table, and a class table with a class table, but a section table cannot be compared with a class table without aggregation or disaggregation, and the method for that change of level must be stated.

How are version and reference period shown in a table?

Version and reference period are shown in the dataset metadata, in the table notes, or in the methodology document that accompanies the dataset, and a reader should record all three places where they appear. Eurostat states on its transition page that parallel datasets presenting current statistics broken down by NACE Rev. 2.1 will be published for many domains, and that backcasting methods will be applied so that previous statistics can be presented by NACE Rev. 2.1. A table in transition may therefore have two versions for the same reference period, and the metadata is the record that distinguishes them. The Central Statistics Office states that its outputs will begin to be published using NACE Rev. 2.1 from January 2026 onwards (FIXED, as stated by the Central Statistics Office), on a gradual basis, starting with Business Statistics and Labour Force Statistics, amongst others. A table from that office must therefore be checked output by output, and the version stated in the output must be recorded, rather than assumed from the office or the year.

What is a break in series, and how is it shown?

A break in series is a jump in a series caused by a classification change, a change in method, or a change in coverage, rather than by a change in the underlying activity. It is shown by a flag, a note, or a break in the line in a chart, in the dataset metadata or the table notes, and it should be quoted with any figure it qualifies. The Eurostat backcasting manual, in its 2025 edition, gives the methods for producing consistent historical series, and it discusses the reconciliation of potential breaks in backcasted series. A backcast series may still carry a break flag where the method cannot fully bridge the change, and that flag is information, not a fault. A reader who quotes a figure across a break without quoting the flag will present a classification effect as a trend, and the quotation will mislead. A reader who quotes the flag with the figure lets the next reader see the break and judge the comparison accordingly. The general range, from 1 January 2025 to 1 January 2031 (FIXED range), frames the transition, but the break for a single dataset is read from that dataset’s notes, not from the range.

Table element	What to record
Heading level	Section, division, group, class, or named aggregation
Classification version	NACE Rev. 2, NACE Rev. 2.1, or the national version, as stated
Reference period	Period the figures describe
Statistical unit	Enterprise, establishment, or other unit, as stated
Break or backcast note	Flag, note, or method, quoted with the figure

The table is a citation checklist for a dataset table. A citation that answers all five rows can be repeated. A citation that gives only the code heading and the figure cannot, because the same heading can appear in two versions with different content.

What does a dataset table not prove about a business?

A dataset table does not prove anything about a named business. A cell total combines many units, and a published cell may be suppressed where it would disclose a single unit, or may be an estimate where direct observation was not available. A code heading in a table is a category label, and the figure in the cell is an aggregate for that category. A reader who treats a table cell as a register entry for a business confuses an aggregate with a record, and the confusion is the same as treating a section total as a statement about one factory. The entry for a specific business is confirmed with the register that holds the record for that unit, not with a dataset table.

Where is a dataset table checked?

A dataset table is checked in its metadata, its table notes, and its methodology document, published by the institute that produced it. The meaning of a heading is checked in the Eurostat explanatory notes. The move between versions is checked in the correspondence tables. The entry for a specific business is checked in the register that holds it. Record the level, version, reference period, unit and break note with every table figure, and confirm a business entry with the register that holds the record.