Field entry, published 2026-11-03, updated 2026-11-03. Group: ISCO-08 in statistics and surveys

How are gender and occupation statistics built with ISCO-08?

Gender and occupation statistics are built with ISCO-08 by taking the occupation group already coded from a person’s described work, and crossing it with the gender information recorded for that person in the same survey or census, so that each occupation group can be shown separately for each gender category used by the source. The construction sounds simple, and in mechanical terms it is, since it is a cross tabulation of two variables. The interpretation is less simple, because the resulting pattern reflects history, education, employer practice, working time, care responsibilities and many other forces, none of which the cross tabulation itself can separate.

What does the cross tabulation actually count?

At its core, the table counts employed people, or jobs, depending on the source definition, within each ISCO-08 group, split by gender. A reader can therefore see, for a given group, how the count divides, and can compare that division with the division for total employment. A group where one gender accounts for a much larger share than in total employment is often described as segregated in occupational terms, while a group whose division resembles total employment is described as more balanced. Those words describe distribution, not causes, and keeping description and explanation apart is the first discipline in reading such tables.

Detail level changes the story considerably. A major group can look relatively balanced while its detailed groups inside it are sharply divided, because opposite imbalances can cancel out in the broad total. Conversely, a detailed group can look sharply divided while resting on a small count or on coding at a boundary. For that reason, gender and occupation analysis usually moves between levels, checking whether a broad pattern survives at detailed level, and whether a detailed pattern is large enough and stable enough to support commentary.

How are the underlying variables defined?

Gender information in official statistics follows the definitions and categories of the source, which vary by country, period and collection mode, and those definitions should be read alongside any table. Some sources record sex, some record gender identity, and some have changed categories over time, so a time comparison must check whether the gender variable itself changed, not only whether the occupation classification changed. A change in categories can move counts between groups in ways that have nothing to do with occupational change.

Occupation coding must also be checked for gender related bias. If job descriptions are recorded differently for similar work, or if coders bring assumptions about who does which kind of work, then coding can reinforce the pattern it is supposed to measure. Statistical offices address this risk through task based group descriptions, coder training and consistency checks, but the risk cannot be eliminated by assertion. Readers should therefore treat gender and occupation tables as measurements that need methods notes, not as transparent windows onto an unchanging reality.

What further breakdowns help, and what do they hide?

Occupation and gender tables gain depth when they are broken down further, by working time, industry, age, education, region or contract type, because those breakdowns can show whether an occupational pattern is accompanied by differences in hours, sector or career stage. For example, a group that looks divided in headcount terms may look different when counted in full time equivalents, if working time patterns differ, and a group that looks divided overall may look different within younger or older age bands. Each further breakdown narrows the question, and that narrowing is useful, provided the resulting cells remain large enough to interpret.

Further breakdowns also carry a warning. Small cells become unstable, disclosure rules may suppress them, and a pattern found in one small subgroup should not be generalised to the whole occupation group. The temptation to chase ever finer breakdowns until a striking pattern appears should be resisted, since fine breakdowns multiply the chance of finding a pattern by chance or by coding artefact. A pattern that matters should be visible at a reasonable level of detail, stable across nearby periods, and consistent with related breakdowns.

How should change over time be read?

Change in a gender and occupation pattern can come from several sources at once: people entering or leaving a group, hours changing, coding changing, the gender variable changing, or the population covered changing. A table that shows movement without those notes invites a simple story about progress or regression that the data may not support. A careful reading asks first whether the classification, the population and the gender categories stayed comparable, then whether the movement appears at both broad and detailed levels, and only then whether a substantive explanation fits the pattern.

When those checks are done, gender and occupation statistics built with ISCO-08 become a valuable descriptive tool. They can show where occupational segregation is strongest, where it is weakening, where it persists at detailed level despite broad change, and where further breakdowns are needed before any explanation is offered. They cannot, on their own, prove why segregation exists, whether it is freely chosen, constrained or discriminatory, or what response would change it. Those judgements need additional evidence about education, hiring, pay, working time and lived experience, linked to the same occupation groups with the same care.

Sources checked

1. ILO: ISCO-08

Back to the ISCO-08 in statistics and surveys group or browse all occupations.