General

A survey in 1978 of 612 serial records in the OCLC data base showed that23% were found to have at least 1 substantial error. 12.6% had only 1 error, 4.4% had 2 errors, 2.6% had 3 errors, and 3.4% had more than 3 errors.                         (Source)

        Ibid…. showed thatthere were errors in the following fields (total is over 100% due to multiple errors):

                title field (field 245, 246, 247)                                                  6.3% errors

                subject heading field                                                               12.0% errors

                imprint field (260)                                                                   19.1% errors

                dates of publication and volume designation (field 362)            45.0% errors

                dates                                                                                      26.0% errors

                succeeding title (field 785)                                                      19.1% errors   

                publication status                                                                    20.5% errors                (Source)

        Ibid…. showed thatthe percentage of times data was omitted in the 6 basic fields was as follows:

                imprint (field 260)                                                                   1.5%

                collation (field 300)                                                               11.7%

                subject heading field                                                              24.5%

                dates of publication and volume designation (field 362)           28.9%

                ISSN (field 022)                                                                    33.5%

                call number (field 050)                                                            35.1%

Only 62.7% had both a call number and subject headings in LC form. Overall, 68% of the records lacked 1 of the basic elements, 14% lacked 2, and 20.4% lacked 3 or more of the basic elements.  (Source)

        Ibid…. showed thatthe percentage of errors in the following fields was negligible, less than 1%: 

                LC call number (050)

                main entry—corporate name (110)

                collation (300)

                key title (222)

                current frequency (310)

                former frequency (321)

                added entry—corporate name (710)

                added entry—conference or meeting (711).                    (Source)

        Ibid…. showed that, of the records with errors, 22% were LC cataloging, 57.4% were non-LC cataloging, and 20% were of “unknown” origin.                 (Source)

        Ibid…. showed that32.5% of the records lacking at least 1 of 6 basic elements (call number in LC form, subject heading in LC form, imprint, collation, ISSN, and date of publication and volume designation) had been authenticated by LC.                     (Source)

A survey in 1978 of 93 serial titles (which had been recently changed), taken from New Serial Tides and compared to title listings in the OCLC data base, showed thatOCLC listed 59.1% correctly and up to date, 17.2% were listed incorrectly, and 23.6% were listed incompletely.                (Source)

A study reported in 1978 of selected fields of 700 non-MARC (i.e., member cataloging) records from the OCLC data base (input after September 1, 1975) showed that 417 or 60% were correct while 283 records contained a total of 393 errors.                         (Source)

        Ibid…. showed thatthe 5 most error-prone fields were: collation (111 or 39% of the total errors), subject headings (88 or 31% of total errors), series (55 or 19% of total errors), added entries (44 or 16% of total errors), and main entry (44 or 16% of the total errors).                       (Source)

A survey reported in 1982 of 144 libraries contracting for OCLC services through the Bibliographic Center for Research (126 or 87.5% responding) showed thatwhen 2 or more copies of a work were acquired at 1 time 44.4% of the respondents reported they would not indicate multiple copy ownership in the OCLC record if all copies went into the same collection, while 30.2% reported they would not indicate multiple copy ownership even if copies went into different collections.               (Source)

        Ibid…. showed thatwhen a subsequent copy of a title cataloged earlier on OCLC was purchased 70.6% of the respondents reported they would not enter information on the subsequent copy into the OCLC record if the copy were going into the same collection as the earlier copy, while 30.2% reported they would not enter information on the subsequent copy even if it were going into a different collection from the earlier copy.                       (Source)

        Ibid…. showed thatwhen the only copy of a work in the library was withdrawn 70.6% of the respondents reported canceling the holdings recorded in the OCLC data base, while 19.8% reported they did not, 6.4% reported varying practices, and 3.2% did not reply to the question.                      (Source)

        Ibid…. showed thatwhen 1 of several copies of a work in the library that were previously cataloged on OCLC was withdrawn 21.4% of the respondents reported that the OCLC holdings were updated, while 65.9% of the respondents reported that the holdings were not, 4.8% reported that their practice varied, and 7.9% did not answer.                         (Source)

Academic

A survey in 1978 of 612 serial records in the OCLC data base showed that32.5% of the records lacking at least 1 of the 6 basic elements (call number in LC form, imprint, collation, ISSN, dates of publication and volume designation) had been authenticated by LC.                 (Source)

        Ibid…. showed thatthe percentage of times data was omitted in the 6 basic fields was as follows:

                imprint (field 260)                                                                  1.5%

                collation (field 300)                                                              11.7%

                subject heading field                                                            24.5%

                dates of publication and volume designation (field 362)         28.9%

                ISSN (field 022)                                                                  33.5%

                call number (field 050)                                                          35.1%

Only 62.7% had both a call number and subject headings in LC form. Overall, 68% of the records lacked 1 of the basic elements, 14% lacked 2, and 20.4% lacked 3 or more of the basic elements.   (Source)

        Ibid…. showed that29% of the records with errors in them had been authenticated by the Library of Congress.                        (Source)

        Ibid…. showed thatthere were errors in the following fields (total is over 100% due to multiple errors):

                title field (field 245, 246, 247)                                                  6.3% errors

                subject heading field                                                              12.0% errors

                imprint field (260)                                                                  19.1% errors

                dates of publication and volume designation (field 362)           45.0% errors

                dates                                                                                     26.0% errors

                succeeding title (field 785)                                                     19.1% errors

                publication status                                                                   20.5% errors                (Source)

        Ibid…. showed thatthe percentage of errors in the following fields was negligible (i.e., less than 1%):

                LC call number (field 050)

                main entry—corporate name (110)

                collation (field 300)

                key title (field 222)

                current frequency (field 310)

                former frequency (field 321)

                added entry—corporate name (field 710)

                added entry—conference or meeting (field 711)                (Source)

        Ibid…. showed that, of the records with errors, 22% were LC cataloging, 57.4% were non-LC cataloging, and 20% were of “unknown origin.”                 (Source)

        Ibid…. showed that23% were found to have at least 1 substantial error. 12.5% had only 1 error, 4.4% had 2 errors, 2.6% had 3 errors, and 3.5% had more than 3 errors.                         (Source)

A survey of 93 recently changed serial titles taken from New Serials Titles and compared to title listings in the OCLC data base in 1978 showed thatOCLC listed 59.1% correctly and up to date, 17.2% were listed incorrectly, and 23.6% were listed incompletely.                    (Source)

A 1979 survey of academic libraries listed in the 1979 edition of OCLC Participating Libraries Arranged by Network and Institution (survey size: 200 libraries; responding: 166 or 83%) showed that, of 109 respondents, 137 (86.2%) reported that they always made substantial checks (i.e., checking “for more than typographical and tagging errors and completing the fixed field”) in at least some types of non-LC records in the OCLC data base. Only 22 (13.8%) libraries reported that they never substantially checked non-LC records. The 3 most frequently checked types of non-LC records were:

                all types                                                              75 (68.8%) libraries

                records from certain cataloging libraries               21 (19.3%) libraries

                audiovisual materials                                            11 (10.1%) libraries                (Source)

A 1979 study at Memphis State University concerning errors in the OCLC data base and based on all error reports generated at Memphis State over a 2-month period (175 error reports) showed that, of the 993 member library records used for cataloging during the test period, 136 (13.7%) contained errors, while of the 2,435 MARC records used for cataloging during this same period, only 39 (1.6%) contained errors.                  (Source)

        Ibid…. showed that, of 175 errors in the OCLC data base that were reported, 28 (16%) were corrected by OCLC within 2 months, and a total of 92 (52.6%) were corrected by OCLC within 4 months.                       (Source)

        Ibid…. showed that, of 93 indexing errors found in OCLC member records, the 4 most frequently reported were: ISBN for monographs (31 errors), LCCN for monographs (22 errors), uniform title for [music] scores (17 errors), and title for [music] scores (10 errors). Of 20 indexing errors found in OCLC MARC records, the most common was ISBN for monographs (19 errors). (Source)

        Ibid…. showed thatthe number of duplicate records found (201 records for 94 titles) was greater than the number of records containing errors (175 records). Specifically, 84 titles had 2 records each, 8 titles had 3 records each, 1 title had 4 records, and 1 title had 5 records.                       (Source)

        Ibid…. showed thatthe cost (labor only) of reporting the 175 OCLC data base errors was $184.07 for 175 error reports.                       (Source)

Special

A study reported in 1983 of the cataloging records for music, music scores, and musical sound recordings over a 2-month period in 1980 from the Indiana University Music Library and the music cataloging unit of the University of Illinois-Urbana Library (996 cataloging records) showed thatthe number of corrections and changes that had to be made in the OCLC record showed wide variation by format. The average book required 6.1 changes per record, the average score required 10.5 changes per record, and the average sound recording required 15.9 changes per record.       (Source)

        Ibid…. showed thatmost of the changes required were in the control data of the OCLC records (control data elements are represented in the MARC record by the OXX fields and are generally more important for use in an online catalog than in card production). Specifically, 79% of the changes required in book records, 71% of the changes required by scores, and 54% of the changes required by sound recordings were control data elements.               (Source)

        Ibid…. showed thatfewer changes were required for cataloging records generated by the Library of Congress than for cataloging records generated by OCLC members. For example, books required an average of 5.5 changes per record for LC cataloging compared to an average of 11.8 changes per record for OCLC members’ cataloging.                 (Source)

 

Dr. David Kohl

 "Libraries in the digital age are experiencing the most profound transformation since ancient Mesopotamian scribes first began gathering and organizing cuneiform tablets."

Go to top