What happens when a level is not split?

When a level is not split in ISIC Rev. 5, UN Statistics Division, the category at the higher level simply has no finer breakdown beneath it at that point in the hierarchy. That is a normal outcome where an activity field is small, uniform or not yet in need of finer detail for statistical purposes. A reader should not invent a subdivision to fill the gap, and should report the category at the level at which it exists.

What does an unsplit level look like in a published table?

An unsplit level looks, at first glance, like something missing. A division appears with no groups beneath it, or a group appears with no classes beneath it, and a reader accustomed to four levels everywhere may search the table for the finer breakdown that is not there. The search will fail, because there is nothing to find. The category at the higher level simply has no finer breakdown beneath it at that point in the hierarchy, and the table is complete as it stands. Reporting the category at the level at which it exists is not a partial answer. It is the full answer the international structure provides for that branch.

Writers sometimes try to repair the apparent gap by inventing a subdivision, borrowing a finer breakdown from a national variant, or repeating the higher level name at a lower level as if repetition created detail. Each of those repairs corrupts the table in a different way. An invented subdivision cannot be checked against the UNSD structure. A borrowed national subdivision describes a different structure, and mixing it with international levels without saying so creates a false comparison. A repeated name pretends that detail exists where it does not. The honest presentation is plainer: show the category once, at its own level, and let the absence of finer rows speak for itself.

Why would designers leave a field unsplit?

A field may be small, so finer detail would split a modest total into fragments too small to publish reliably. It may be uniform, so the work inside it does not divide naturally into distinct kinds that statistical users need to separate. Or it may not yet be in need of finer detail for statistical purposes, because users have not asked for the breakdown, or because the breakdown would not change any decision the tables are meant to inform. In each case, leaving the level unsplit is a judgement about usefulness, not an oversight. Detail that nobody needs, and that the evidence cannot support, adds noise rather than information.

That judgement can change. A field that is small today may grow, a uniform field may diversify as a new method of production spreads, and users may begin to ask for a breakdown that once seemed unnecessary. A future revision may then split the level, and a category that once stood alone may acquire groups or classes beneath it. Until that revision happens, however, the unsplit level is the structure, and a reader should work with it as it stands rather than anticipating a split that has not yet been published. Anticipation is a form of invention, and invention has no place in a table that claims to follow Rev. 5.

What should a writer do instead of filling the gap?

Report at the level that exists, and say so. A short methods note can carry the whole point in one sentence: the category is reported at division level, for example, because Rev. 5 provides no finer breakdown for that branch. That sentence prevents a later reader from assuming the writer was careless, or from searching for detail that was never published. It also protects the writer, because a stated level can be checked, while an invented subdivision can only be challenged.

The working habit behind that sentence is the same as elsewhere in this cluster. Name the activity in words. Find the section and division in the UNSD Rev. 5 structure. Read the group and class notes where they exist, and note where they do not. Record ISIC Rev. 5, UN Statistics Division, beside the result, so the version can be checked later. Work through those steps in order, and do not skip the note reading step, because that step is where borderline cases are decided, and where an unsplit level is confirmed rather than assumed. A level that has been checked and found unsplit is a finding. A level that was never checked is only a gap in the writer’s notes.

This guide describes a feature of a statistical classification only, not a ruling on any reader’s affairs. On an unsplit level, this guide stays within statistical classification only and fixes no reader’s code, principal activity, or registration; a specific record is checked with the national statistical institute or business register holding it.