The quality of the member-input monographic records in the OCLC on-line union catalog

James Gerald Schoenung

doi:10.17918/00009309

The past decade has witnessed a revolution in libraries brought on by the emergence of computerized library networks. The OCLC Online Computer Library Center, Inc., the largest of several bibliographic utilities providing computer-based library services, has made large-scale cooperative cataloging a reality for over 2,400 institutions. At the same time, it has served to spotlight long-standing problems associated with shared cataloging, such as the accuracy, consistency, and completeness of cataloging data supplied by heterogeneous libraries having varying bibliographic requirements. The purpose of this research was to evaluate the effectiveness of the OCLC Shared Cataloging Subsystem, with particular emphasis on the quality of member-input monographic catalog records. The methodology was to take a random sample of 1,107 participant-supplied records added to the data base between 1971 and 1977 and examine them against a predetermined standard of cataloging having wide acceptability. Member-input records were compared against the same cataloging produced by the Library of Congress. Deviations between a sample record and its corresponding LC copy were identified as "errors," as were failures to follow OCLC input standards. A large portion of the variations between member-input and LC catalog records were not really errors incurred by the inputting library but rather resulted from changes in OCLC input standards over the years. A multifaceted error categorization instrument was used to code errors according to location, type, cause, and effect. Variables characterizing sample records and inputting libraries were also collected and related to the frequency and types of errors. Errors and variations from LC copy were coded in such a way as to be grouped into one of four error effect categories, depending on the seriousness of the error: (1) those affecting access points (fatal), (2) those involving duplicate records or affecting a work's description at the edition level (serious), (3) those which are neither fatal nor serious (minor), and those variations from LC copy resulting from differences in style rather than inaccurate information (other). In all, 16,186 errors and variations from LC copy were identified in the 1,107 sample records, with the average member-input record containing 14.6 errors. Only nine of the sample records were without any errors and the variation of errors within records was substantial, with errors tending to "clump" in certain records. The average record had 0.62 fatal errors, 0.32 serious errors, 10.94 minor errors-of which 7.11 occurred in fixed field data and 3.82 in variable fields, and 2.74 other errors. Errors were about evenly divided between fixed field data elements (7,875) and variable fields (8,311). Approximately 2.6% of the member-input records were cataloged on incorrect bibliographic formats. The most visible deficiency of the OCLC data base is the high incidence of duplicate records. It is estimated that as many as one in six of the monographic records is in fact a duplicate. Findings relating to access indicate that, excluding subject searching which is not available in OCLC, data base access is in most ways superior to that available in conventional card catalogs. Of all measured variables, the one most strongly associated with catalog record quality is the year of sample record entry into the data base.

The quality of the member-input monographic records in the OCLC on-line union catalog

Files and links (1)

Abstract

Metrics

Details

The quality of the member-input monographic records in the OCLC on-line union catalog

Files and links (1)

Abstract

Metrics

Details

Drexel University Social media