What makes categories exhaustive and mutually exclusive?

Categories in a statistical classification are exhaustive, mutually exclusive, and allocated by consistent methodological principles. The Eurostat manual states those three characteristics together: exhaustive coverage of the observed universe, mutually exclusive categories where each element must be classified in only one category, and methodological principles which support the consistent allocation of elements to the various categories. The background article states the same characteristics, and adds that classifications allow statistics to be organised by meaningful categories, defined so that the universe of observations is partitioned into sets that are as internally homogeneous as possible with respect to the characteristic of interest.

What does exhaustive coverage mean?

Exhaustive coverage means that each element in the observed universe can be classified in one of the categories. The manual states the point in the form of a requirement at each level: the categories at each level must be mutually exclusive and collectively exhaustive of all objects in the population of interest. At a given level, the categories together must cover the whole field, with no activity left without a place. The residual categories marked n.e.c., and headings including the term other, are the mechanism that secures exhaustive coverage inside a parent. The manual states that the scope of a position is required to be exhausted by the positions below it, and residual categories gather the remainder.

Exhaustive coverage does not mean that every activity has a specific named category of its own. It means that every activity has a place, specific or residual, inside the hierarchy. An activity that is not named in a specific class note can still be placed, if it falls inside the parent scope and outside the specific categories, in the residual category for that parent, subject to the residual note. A reader who cannot find a named category for an activity must therefore read the parent note and the residual note, rather than concluding that the activity has no place in the classification. The place may be residual, and the note must be read to confirm it.

Characteristic	Meaning stated in the manual	Mechanism in NACE
Exhaustive coverage	Each element can be classified in one category	Residual n.e.c. and other categories complete each parent
Mutually exclusive	Each element is classified in only one category	Notes state inclusions and exclusions with cross references
Consistent allocation	Methodological principles support consistent allocation	Classification rules in the introductory guidelines

What does mutually exclusive mean?

Mutually exclusive means that each element is classified in only one category at a given level. The manual states the point directly: each element must be classified in only one category of the classification. At class level, that means one class for the principal activity of a unit in a statistical register, after all activities have been considered. At higher levels, it means that the categories at that level do not overlap, so that aggregating the categories at a level gives the total for the parent without double counting. The hierarchy supports the rule: each lower category belongs to exactly one category above, so that a class counted in one group is not also counted in another group.

The explanatory notes support mutual exclusivity through exclusions and cross references. The manual states that notes highlight boundary issues by providing examples of activities that may appear similar to those belonging to a position but are classified elsewhere. An exclusion in one note, with a cross reference to another class, is a statement that the activity belongs in only one of the two classes, and the cross reference states which one. A reader who reads only the inclusions in one note, without reading the exclusions, may place an activity in two classes at once. Reading both lists, and following the cross references, preserves mutual exclusivity.

What are the methodological principles for consistent allocation?

The methodological principles are the classification rules in the introductory guidelines. The manual states that the categories are defined in terms of characteristics of a population of units of observation, and that hierarchical classifications are characterised by an increasingly granular partition of categories, which makes it possible to collect and present information at various levels of aggregation. The rules then state how a unit with several activities is assigned: the principal activity is the activity contributing most to total value added, value added substitutes are used where value added cannot be obtained, the top down method is used where no activity accounts for more than half of value added, and specific rules apply to outsourcing, intermediation, trade, and other cases.

Consistent allocation means that the same rules, applied to the same information, give the same assignment. The rules do not guarantee that two readers with different information will reach the same assignment, and they do not guarantee that a reader with incomplete information will reach the assignment that the register reaches with full information. Consistency is a property of the method, not a guarantee for an individual reading. A reader who applies the rules as described in the guidelines, without applying them to a specific business, is learning the method. A reader who needs an assignment for a unit must put the full information to the institute or register that holds the record.

What do the three characteristics not guarantee?

The three characteristics do not guarantee that a code is easy to find, that a heading is self explanatory, or that a code decides a legal question. Exhaustive coverage guarantees a place, not a named place. Mutual exclusivity guarantees one place at a level, not that the place is obvious from the heading. Consistent allocation guarantees a method, not that every reader will reach the same result from partial information. The notes, the index, and the correspondence tables are the tools that make the characteristics usable in practice, and the institute or register is the place where an assignment is confirmed.

Read the manual for the three characteristics, read the notes for the boundary statements that make them work, and confirm any entry with the national statistical institute or the business register that holds the record.