The description of statistical tables: a problem in data documentation.

Abstract

Reference to and direct storage of primary data is of growing importance. This explains the interest in the description of statistical tables. The brute-force approach to give each stratifier (that is, the list of all row captions or the list of all column captions) in full has its economical limits. A method is described which makes the description of a large number of large tables economically feasible. The basic idea is to replace the full listing of a stratifier (e.g. US counties) by a reference to a "prefabricated" stratifier available in a classification scheme (e.g. US geographical division) taking into account the hierarchical structure of such a scheme. Modification of prefabricated stratifiers is also possible in order to adapt to small variations occurring in actual tables. The method has also applications in thesaurus building.