Field entry, published 2026-11-03T00:00:00.000Z, updated 2026-11-03T00:00:00.000Z. Group: National variants, labelled as variants

How do you convert a national code to ISCO-08?

You convert a national code to ISCO-08 by finding the official correspondence for the exact national system and version you hold, applying it at a level it supports, and recording the mapping so that a later reader can repeat the step. You do not convert by copying digits, by matching titles that sound alike, or by guessing from a similar group. Conversion is a documented method, not a shortcut, and a national code that has not been through that method remains a national variant code, however it is displayed.

That starting point protects both the original data and the mapped result. The original data keep their national meaning, and the mapped result gains an international meaning only to the extent the correspondence supports it.

What do you need before you start?

You need five pieces of information, and conversion should wait until all five are in hand. First, the name of the national system. Second, its version, because correspondences are version specific. Third, the level of the national code, since a broad national group and a detailed national group do not map in the same way. Fourth, the correspondence itself, from the national body that publishes the variant or from a statistical body that publishes the mapping. Fifth, the ISCO-08 level at which you intend to report, remembering the ILO hierarchy of 10 major groups, 43 sub-major groups, 130 minor groups and 436 unit groups.

If the correspondence cannot be found, stop. An undocumented conversion is worse than no conversion, because it creates a figure that looks international while resting on an unstated judgement. In that case, present the national figure labelled as a national variant, and note that mapping is still required.

What does a careful conversion sequence look like?

A careful sequence runs in order, and each step leaves a trace.

1. Confirm the source system, version and level for every code in the set, not just for a sample, because a mixed version dataset will map inconsistently if treated as uniform.
2. Obtain the correspondence that matches those versions, and check its coverage notes for groups that map one to many, many to one, or not at all.
3. Apply the correspondence at the broadest level where coverage is documented as reliable, and resist the urge to carry mapped detail finer than the correspondence allows.
4. Handle split groups by the rule the correspondence states, whether that is a best fit, a proportional split based on stated auxiliary data, or reporting at a broader level. Do not invent a split rule silently.
5. Set aside unmapped or residual groups openly, rather than forcing them into the nearest ISCO-08 group, and report their size so a reader can judge the impact.
6. Check the mapped totals against the original totals, and investigate any difference before publishing, since a mapping should explain, not create, a change in total.
7. Write the method note before the table is shared, naming the national system, its version, the correspondence version, the ISCO-08 level and any groups handled specially.

That sequence is deliberately slow, and a conversion that cannot be repeated from its method note has not been completed.

How do you handle one to many and many to one cases?

One to many cases arise when one national group spans several ISCO-08 groups. The honest options are to report at a broader ISCO-08 level where the span disappears, to apply a stated split rule, or to flag the group as spanning and exclude it from detailed comparison. Choosing silently among those options hides a judgement that affects the result.

Many to one cases are simpler but still need care. Several national groups collapse into one ISCO-08 group, so detail is lost. The mapped figure is valid at ISCO-08 level, but the national detail should be preserved in the original dataset, because it cannot be recovered from the mapped figure later. Analysts who need that detail must go back to the national coded data.

What should the published result carry with it?

The published result should carry its history in miniature. A table note, a chart footnote or a method paragraph should state that the underlying data were coded in a named national variant, give its version, name the correspondence used, give the ISCO-08 level, and mention any residual or split groups. That note need not be long, but it must be present where the figure is first shown, not only in a separate technical document that most readers will not open.

A conversion that carries that note can be checked, challenged and improved. A conversion without it can only be believed or disbelieved, which is a poor basis for statistical work.

What is not provided by this guide?

This guide provides no correspondence table, no national codes and no version specific mapping rules, because those must come from the current publication of the relevant national body and from the ILO structure for ISCO-08. Any guide that printed a correspondence here would risk going out of date and misleading readers who copied it without checking. Use this guide for the sequence and the checks, and use the primary publications for the entries themselves.

Sources checked

1. ILO, ISCO-08

Back to the National variants, labelled as variants group or browse all occupations.