Metadata for units of measure in social science databases

Abstract.

Statistical social science databases present many unique data management problems. Among them is the fact that computer-aided use of statistical social science data is subject to serious but preventable errors in data manipulation and analysis, particularly given the increasing number of non-expert users. Two incremental prototype solutions to the problem of correctly manipulating units of measure when manipulating data are described. These implementations use rudimentary metadata representing measurement-theoretical and category attribute information to support error checking of data derivations common in research and reference using social science data. The specification of the metadata structures and the implementation of its processing is informed by relevant principles from literature in statistical and scientific data models, dimensional analysis, and knowledge representation with formal ontology. Design parameters useful in the effort to establish appropriate variable-level metadata specifications for social science databases used in digital environments are discussed. The integrated use of the proposed metadata appears to be a promising method of better supporting the manipulation and analysis of social science data. While measurement theoretical metadata has further application to the cataloging and retrieval of collections of statistical databases, the importance is argued of metadata for social science databases to serve users beyond these purposes.