What is a statistical unit in ISIC?

Before an activity can be classified, statisticians must decide what is being classified, and that unit is called a statistical unit in ISIC Rev. 5, UN Statistics Division. A unit might be an enterprise, an establishment or another defined part of an organisation, depending on the statistical programme. The choice matters because the same organisation can carry out several activities in different places, and each unit is classified by its own activity.

Why must the unit be fixed before the activity is coded?

Coding an activity presupposes an answer to a prior question: coding the activity of what? A statistical unit is that answer. In ISIC Rev. 5, UN Statistics Division, the unit might be an enterprise, an establishment or another defined part of an organisation, depending on the statistical programme in use. The programme decides, because different programmes need different grains of observation. A survey of enterprises wants the whole enterprise as its unit. A survey of production at particular sites wants establishments. Another programme may need a still more precisely defined part. None of those choices is wrong in itself, but mixing them without saying so produces counts that cannot be compared, because the things being counted were never the same kind of thing.

The choice bites hardest inside a single organisation. The same organisation can carry out several activities in different places, and each unit is classified by its own activity. Collapse those units into one, and the organisation’s secondary work disappears into its main work, or its work in one place is credited to another. Keep the units distinct, and each activity can be counted where it actually happens. The unit decision is therefore not administrative tidying. It decides what the eventual industry figures describe.

How does a reader tell units apart in a published table?

A published table rarely announces its unit choice in its title, which is why a careful reader looks for it in the methods note. The note should name the programme, the unit, the version and the level, and a reader who cannot find those facts should treat the table with reserve until they are found. Version, level, reference period and source are the four checks that repay the effort here. ISIC Rev. 5, UN Statistics Division, should be named where Rev. 5 is meant, and a different label should never be silently substituted. Level matters because a division total and a class figure answer different questions even when they share a word in their titles. Reference period matters because revisions and national adoption dates can fall inside a series. Source matters because an unnamed classification cannot be checked.

Consider two tables that look alike but rest on different units. One counts enterprises, the other counts establishments, and the second will naturally show more units and a different spread across activities, because a multi-activity, multi-site organisation contributes several establishments but one enterprise. Neither table is mistaken. The mistake would be to compare them as though they counted the same thing, or to quote one as though it answered the other’s question. The methods note is what prevents that mistake, and reading it is part of reading the table.

What happens at the edges, where units blur?

Real organisations resist tidy boxes. An organisation may combine several activities, an activity may change character over time, and a new method of production may blur an old boundary. ISIC Rev. 5, UN Statistics Division, meets those cases with explanatory notes, includes and excludes lists, and residual categories, but those tools only help if they are actually read. A reader facing a borderline case should slow down, read the notes for both candidate categories, and note the uncertainty, rather than forcing a quick answer that later proves hard to defend. The same patience applies to units: where it is unclear whether a part of an organisation counts as a separate unit for the programme in question, the programme’s own definition of a unit, not a general impression, is the authority to consult.

What should a methods note record about units?

• Start from the activity in words, not from a code remembered from an earlier revision.
• Read the explanatory note before settling on a category, especially where two titles look similar.
• Keep the version label, ISIC Rev. 5, UN Statistics Division, with any code that leaves your notes.
• Where detail cannot be justified, report at the broader level that is clearly correct.
• Note any break in series when figures from different revisions appear in one comparison.
• Name the statistical unit, so a reader knows whether enterprises, establishments or another defined part were counted.

Treat the note as part of the result. A figure without its unit, version and level is a figure waiting to be misunderstood.

This is general information about statistical units and classification, not a ruling on any reader’s situation. It does not determine a code, a principal activity or a registration for anyone, and a specific record should be checked with the national statistical institute or business register that holds it.