Why was Rev. 4 Section J split?

Rev. 4 Section J was split in ISIC Rev. 5, UN Statistics Division, because a single information section had come to cover activities that had grown apart in production character. Publishing, broadcasting and content production on one side, and telecommunications, computing and information services on the other, no longer sat comfortably in one broad block. Splitting the section let each block be described at section level in its own terms, rather than sharing a heading that obscured the difference.

What had the old single section come to contain?

By the time of Rev. 5, the old Section J had become a very large tent. Inside it sat publishing, broadcasting, and content production, work whose output is material for an audience, alongside telecommunications, computing, and information services, work whose output is the systems and services around information. Those two families grew side by side for years, and as each grew, the distance between them grew too, in production character, in the skills and equipment involved, and in the kinds of tables readers wanted to build from them. A single section heading had to stretch across both families, and the stretch showed: a reader opening the old section found content work and infrastructure work listed as neighbours, with the heading doing little to explain what, if anything, they shared beyond a broad association with information. The section had not become wrong. It had become too broad to describe its contents usefully at section level.

Why did one heading obscure two different production characters?

A section heading works by telling a reader, at a glance, what kind of production the section contains. When the contents divide into two production characters, a single heading must either favour one side or blur both, and either choice misleads someone. Favouring content would have hidden the infrastructure work. Favouring infrastructure would have hidden the content work. Keeping a neutral, umbrella wording would have preserved the problem that prompted the revision, namely that a reader could not tell from the section alone which family a given activity belonged to, and a table total for the whole section would have mixed two different kinds of production into one number. Splitting the section let each family be described at section level in its own terms: Section J for publishing, broadcasting, content production, and distribution, and Section K for telecommunications, computer programming, consultancy, computing infrastructure, and other information services. Each heading now says something specific, and a reader can test an activity against the right heading before moving to divisions and classes.

How does the split change the way a reader compares old and new tables?

The split changes comparisons in two ways that a reader must keep separate. First, section level totals are no longer directly comparable across the revision, because the old single total has been divided between two new sections, and neither new total alone equals the old one. Second, the later sections of the alphabet shifted letters after the insertion, so a reader comparing section letters across revisions must remap them through the correspondence material rather than carrying letters across unchanged. A careful comparison therefore starts by writing each block in words, using its own revision’s definitions, then aligning the blocks through the correspondence table, and aggregating to a level where the blocks line up where a fine alignment is not possible. The published comparison should flag the break at the revision point, name the version, ISIC Rev. 5, UN Statistics Division, state the level, and give the reference period, so readers can see what changed in the classification and what changed in the world being measured.

What habit makes the split useful rather than merely tidier?

The split is useful only if readers use the new headings as a first decision, not as decoration. That means describing the activity in words, deciding at section level whether it belongs with content or with infrastructure, and only then moving down through divisions, groups, and classes, reading the explanatory notes and excludes lists at each level. It means keeping the version label with any code or figure that leaves the reader’s notes, so a later reader does not mistake a Rev. 5 total for a Rev. 4 one. And it means recording uncertainty, and reporting at a broader level where finer detail cannot be justified, rather than forcing a fine category that a checker cannot reproduce. A survey team that updates its questionnaire, coding manual, and published table together, and a reader who checks version, level, period, and source before quoting a figure, both turn the split from a tidier set of headings into a more reliable set of tables, which is what the revision was for.

This guide addresses statistical classification only, and it does not decide anything for a particular reader: it does not determine any reader’s code, principal activity, or registration, and confirmation for a specific record should be sought from the national statistical institute or business register that holds the record.