A design for a census information retrieval system

This article explores the methodology of designing a storage and retrieval system for information on the nature and subjects of censuses published internationally over the past two decades. The essential key is neither data storage nor bibliography but an index which will tell a researcher what censuses provide information on particular population characteristics and how those characteristics are defined, mapped, cross-indexed, compared, and discussed.The index should be based on specific, inclusive, and flexible classifying terms. The authors arrive at 24 variables which are grouped on two levels-category and specific item-by the use of either main tables or indexes. The use of indexes is preferred for its reduction of clerical and computer time.In addition to the index (which should be made available in machine-readable form), the authors suggest the concomitant publication (on the basis of unified classifications) of an analysis of the relative value of different definitions of population characteristics, a merged-item vocabulary, and indexes to various censuses. Once the study has developed beyond an experimental phase limited in geographic extent and in time, the authors envision a continuing service which will provide indexes to all new censuses.