What is a division in ISIC Rev. 5?

A division in ISIC Rev. 5, UN Statistics Division, is the second level of the hierarchy, numbered with two digits, and Rev. 5 contains 87 divisions in total. Divisions break a section into recognisable fields of activity, finer than a section but still broad enough for many published tables. A reader who wants a sector overview will often stop at division level, while a reader who needs detail will go further down.

What makes a division the right stopping point for a sector overview?

A division is the second level of the hierarchy, numbered with two digits, and Rev. 5 contains 87 divisions in total. Divisions break a section into recognisable fields of activity: finer than a section, but still broad enough for many published tables. That middle position is their strength for an overview. A section total can feel too coarse, hiding the differences between fields that a reader of a sector overview actually wants to see. A class figure can feel too fine, burying the overview in distinctions that belong in a specialist table. A division total sits between those faults, showing fields a reader can recognise and compare, without demanding specialist knowledge of the finer breakdown.

A writer preparing a sector overview should therefore begin at division level and justify any move away from it. Moving down to group or class level needs a reason: a question about variation inside the field, or evidence that supports the finer breakdown. Moving up to section level needs a reason too: a question about the broad block as a whole, where field differences would distract. Without such a reason, the division total is the default, not because it is a compromise, but because it matches the scale of the question most overviews actually ask.

How does a division hold together fields that later split apart?

Inside a division, groups split the field into clusters of related activities that share a production character, and classes split those groups into the finest detail the international classification offers. The division total therefore contains a whole internal structure, even when the published table shows only the total. A reader who wants to understand the total can drill down through its groups and classes, checking how the field is composed, while a reader who only needs the overview can stop at the total and still trust it, because the total is the sum of those finer parts and nothing else.

That containment also explains why a division total and a class figure can share a word in their titles yet answer different questions. The division total aggregates every group and class beneath it, while the class figure isolates one kind of work inside one group. A writer who quotes one while meaning the other will mislead, even if the numbers are both correct in their own terms. Naming the level beside every figure, division or class, prevents that swap, and a table heading that states the level plainly is worth more than a footnote that a hurried reader may miss.

How should a writer move between division and finer levels in one article?

Move deliberately, and signal each move. Open with the division total as the overview, then, where a question needs it, drill down to one or two groups, explaining in words why those groups matter for the question at hand. Return to the division total before moving to the next field, so the reader never loses the overview while following the detail. That rhythm, broad, then narrow, then broad again, mirrors the hierarchy itself, and it lets a reader who only wants the overview skip the detail without losing the thread.

In working notes, the same discipline applies at the level of the coder and the analyst. A survey questionnaire, a coding manual and a published table should all point to the same category definitions, or results will drift between collection and publication. ISIC Rev. 5, UN Statistics Division, provides that shared reference, so the coder, the analyst and the reader work from the same structure, and the version label should travel with the data at every stage, from the first coding decision to the final chart. A lost label is how cross revision errors enter an otherwise careful analysis, and a division total without its label is the easiest place for such an error to hide.

This guide is about one level of a statistical classification, and it offers no verdict on any reader’s own position. Nothing in this guide on a division determines a reader’s code, principal activity, or registration, since it is statistical classification only; the national statistical institute or business register holding a specific record confirms it.