How do the four levels nest?

The four levels of ISIC Rev. 5, UN Statistics Division, nest like boxes inside boxes: every class belongs to exactly one group, every group to exactly one division, and every division to exactly one section. That nesting is what lets a statistician aggregate detailed data up to a broad total without losing the audit trail, and lets a reader drill down from a headline figure to its components. If a total does not equal the sum of its parts, the nesting is the first place to check for a version or grouping error.

How does one class find its single home in the hierarchy?

Nesting in Rev. 5 is strict in one direction. A class sits inside one group only, that group sits inside one division only, and that division sits inside one section only. No class is shared between two groups, and no group is split across two divisions. That single parent rule is what makes aggregation trustworthy. When a statistician adds class figures to reach a group total, then group figures to reach a division total, then division figures to reach a section total, every detailed figure is counted once, in exactly one place. A reader who understands that chain can move in either direction with confidence: downward from a headline to see what it contains, or upward from detail to see where it belongs.

The practical value shows up when a total looks wrong. Suppose a published section total does not match the sum of the divisions shown beneath it. The nesting rule gives the reader a first diagnostic question, not a verdict: has a division been left out, has a figure been taken from a different level, or has a figure from another revision slipped into the table? Each of those faults breaks the chain in a different place, and each is easier to spot once the reader expects every part to belong to one parent only. A total that cannot be retraced through its parents is not yet ready to quote.

Why does adding a level change what a total can show?

Each step down the hierarchy narrows the question being answered. A section total describes a large block of the economy in broad terms. A division total describes a recognisable field inside that block. A group total describes a cluster of related activities inside the field. A class figure describes the finest detail the international classification offers. Those are not four ways of saying the same thing at different lengths. They are four different questions, and a reader who swaps one for another will draw conclusions the data does not support.

Consider a writer preparing a short article about a broad economic block. A section figure may be the right headline, because the article is about scale and comparison at the broadest level. The same writer, asked a week later to explain how work inside one field differs, would need division or group figures, because the section total hides exactly the variation the new question asks about. Moving down is not automatically better. Detail helps only when the question needs it and the evidence supports it. Moving up is not automatically vaguer either. A broader total can be the more honest figure when the finer breakdown cannot be justified from the material available.

How would a reader retrace a headline total to its parts?

Start with the headline and ask, level by level, what it contains. Which divisions make up the section? Which groups make up each division? Which classes make up each group? At each step, check the UNSD Rev. 5 structure rather than memory, because memory preserves the shape of an earlier revision or a national variant. If a step cannot be answered from the structure, stop there and report at the broader level that is clearly correct, noting the gap instead of filling it with a guess. That pause is not a failure of the method. It is the method working as intended, protecting a broad figure from detail it cannot defend.

A table built this way reads differently from a table built from remembered labels. Its totals can be checked, its parts can be challenged, and its limits are visible. A reader in another country, opening the same UNSD structure, can follow the same chain and reach the same understanding of what each total contains. That shared retracing is the point of a nested hierarchy: not elegance for its own sake, but a structure that lets strangers check each other’s arithmetic.

This guide explains a statistical classification only, and it settles nothing about any individual reader. It does not fix a reader’s code, principal activity, or registration, and any code for a particular record should be confirmed with the national statistical institute or business register that holds that record.