Content area
Full Text
Better quality control over the subject access points in cataloging records is essential if we are to improve retrieval effectiveness or to facilitate cooperative cataloging. An important measure of catalog quality is how consistently individual records are indexed. Findings of earlier inter-indexer consistency studies indicate that there is a direct correlation between retrieval effectiveness and inter-indexer consistency; that is, high inter-indexer consistency in assignment of indexing terms appears to be associated with a high retrieval effectiveness of the documents indexed.
Most MARC records are prepared according to widely accepted standard tools such as Anglo-American Cataloguing Rules, second edition, (AACR2), Library of Congress Subject Headings (LCSH), Dewey Decimal Classification, and Library of Congress Classification. When these tools are carefully followed, the result is what we accept as standard practice, which, for want of an absolute measure, we define as congruence with Library of Congress (LC) cataloging. Some discrepancy in record treatment is inevitable, given the variance in the nature of the material cataloged and the degree of subjective judgment required to use the tools. We find justifiable inter-cataloger differences even in descriptive cataloging, where circumstances are more clear-cut than in subject cataloging. Discrepancy is naturally considerably higher for subject cataloging, where guidelines are less distinct and where there is much more scope for differences in assessing the material. Still, it must be our goal to reduce inter-indexer inconsistency as far as possible. To do so, we need more information than we have on the inconsistencies we all know exist.
The problem has been addressed in numerous studies in the past.' Many of the earlier studies, however, suffer from the lack of reliable means of measurement because of the presence of uncontrolled variables. Some have relied on samples prepared specifically for the experiment, having the same documents indexed by two or more indexers for comparison, resulting in consciously or artificially prepared samples. Others have compared records extracted from similar databases but possibly indexed according to different policies. Clearly, a more valid way of measuring inter-indexer consistency is desirable. The work reported here details a new approach to measuring inter-indexer agreement on subject heading assignment.
The purpose of the current study has been twofold: first, to develop a valid methodology for studying indexing consistency in MARC records and,...