Field entry, published 2026-11-03, updated 2026-11-03. Group: ISCO-08 in statistics and surveys
How are employment statistics grouped with ISCO-08?
| Major group | All major groups, ISCO-08, ILO |
|---|---|
| Skill level | See text, ISCO-08, ILO |
| Code ladder | ISCO-08 moves from a broad major group down through sub-major and minor groups to a detailed unit group. |
Classification only, not a job decision
This guide is about grouping employment counts with ISCO-08 for statistical tables, and it does not classify the work of any individual reader.
Employment statistics are grouped with ISCO-08 by counting jobs at a detailed level first, then adding those counts upward through the classification hierarchy whenever a broader table is needed. A detailed count sits in one unit group, and that unit group belongs to exactly one minor group, which belongs to exactly one sub-major group, which belongs to exactly one major group. Because the hierarchy nests cleanly, a broad total can be built by summing its parts, and a reader can move between levels without double counting, provided the underlying coding was consistent. Grouping is therefore an aggregation exercise built on coding, not a separate act of invention at each level.
What does each level add, and what does it hide?
| Level in the hierarchy | What it offers a reader |
|---|---|
| Major group | A broad view of the occupational structure, useful for headline comparisons |
| Sub-major group | A finer split that separates broad families into recognisable fields |
| Minor group | A more detailed view for studying particular kinds of work |
| Unit group | The most detailed ISCO-08 view, closest to the tasks people described |
Moving up the hierarchy gains stability and comparability, because small coding differences at detailed level often wash out in a broader total. Moving down gains specificity, because differences that disappear in a major group total become visible again. A good statistical release offers both, with notes explaining when a detailed cell rests on limited material.
How do statistical offices decide how much detail to publish?
Publication detail depends on the source, the quality of the underlying descriptions, and the risk of identifying individuals in small cells. A census can often support more detail than a sample survey, because its counts rest on broader coverage, while a survey may need to collapse detailed groups before publication to keep estimates reliable. Offices also consider user need: a table that is too detailed for its source invites misreading, while a table that is too broad hides the very differences a researcher came to study. The balance is struck case by case, and it should be visible in the table notes.
Vague or incomplete descriptions create a further decision. Some jobs can only be coded at a broader level because the answer did not contain enough detail for a unit group placement. Offices may publish a residual or not elsewhere classified category at that broader level, or they may apply documented rules for allocating such cases. Either approach affects interpretation, and a reader who ignores the residual category may overstate the size of the named groups. Checking the size and treatment of residual categories is a simple habit that prevents many errors.
Why do two employment tables sometimes show different totals for the same occupation?
Different totals usually trace back to different definitions, populations or levels of aggregation rather than to arithmetic mistakes. One table may count employed people, another may count jobs, and a person with two jobs will appear differently in the two concepts. One table may cover a resident population, another may cover people working in a territory, and cross border workers will shift the totals. One table may use a major group total, another may use a single unit group, and the labels can look similar while the contents differ sharply. Reading the table title, the population definition and the classification level together resolves most apparent contradictions.
Timing matters as well. Employment statistics are tied to reference periods, and a table for a single week, a month or a year will capture seasonal work differently. Occupation groups with strong seasonal patterns can therefore look larger or smaller depending on when the data were collected. That variation does not indicate a coding problem. It indicates that employment is a flow through time as well as a stock at a moment, and ISCO-08 grouping must be read with the reference period in mind.
How can grouping be used well, and where does it mislead?
Grouping works well when the question matches the level. Questions about broad structural change suit major groups, questions about a field suit sub-major or minor groups, and questions about a specific kind of work suit unit groups, subject to quality notes. Grouping misleads when a broad label is treated as if it described a single kind of work, or when a detailed label is treated as if it captured every job title in everyday language. ISCO-08 groups are statistical categories defined by tasks and duties, not dictionaries of job titles, and titles that sound alike can sit in different groups when the work differs.
A final caution concerns change over time within a fixed hierarchy. If work itself changes, a group can grow because new tasks enter it, because existing jobs are redescribed, or because coding practice shifts, and a grouped total alone cannot separate those causes. Grouped employment statistics are powerful precisely because they simplify varied working lives into comparable counts, and they stay honest only when readers remember that simplification, check the level, read the population definition, and consult the notes on residuals and coding before drawing a conclusion.
Sources checked
Back to the ISCO-08 in statistics and surveys group or browse all occupations.