USER PERSPECTIVES ON THE PROGRAM FOR COOPERATIVE CATALOGING BIBCO CORE RECORD STANDARD

A Pilot Study of the Core Bibliographic Record For Books

Executive Summary from the Final Report^PDF

Karen M. Letarte
North Carolina State University

Michelle R. Turvey
Southwest Missouri State University
September 30, 2001

A. Introduction

In 1994, the Library of Congress introduced the Program for Cooperative Cataloging (PCC) Core Record Standard. The Core Record Standard is intended to reduce cataloging costs by increasing both the pool of available cataloging records and the speed of cataloging while ensuring an acceptable level of data quality. The program's goal is to create a pool of records that can be used with a "minimum of intervention"^[1] by libraries. However, the use of the Core Record Standard has been somewhat controversial in the library community. PCC Core records provide less information than full-level catalog records but more than minimal-level cataloging records. This has led to the question of how well Core records meet the needs of users, particularly with respect to the sufficiency of bibliographic data. There has been no study to date of the impact of the Core on catalog users. This pilot study's focus is the ability of the Core Bibliographic Record standard for Books to meet the information needs of catalog users.

The pilot study was conducted by Karen M. Letarte and Michelle R. Turvey in March 2001 at Southwest Missouri State University (SMSU). The study was funded by the PCC and through a Faculty Research Grant from Southwest Missouri State University. Additional funding was provided by the Dept. of Library Science at SMSU. Karen Calhoun (Cornell University) and Mike Prasse (OCLC) assisted with the initial design of the research. Anne Schalda, Institutional Researcher (Drury University), was the statistical consultant for the study.

B. Research Design and Methods

A random sample from the SMSU Libraries' patron database formed the pool from which the participant group of 75, comprised of 26 faculty members, 25 undergraduates, and 24 graduate students, originated. Participants were asked to examine 4 catalog records for social science monographs, 2 cataloged at Core level, and 2 cataloged at Full level. The records were chosen from the OCLC database, with Library of Congress cataloging preferred. A copy of each record is included (see Appendix A). Following the specifications for the Core Bibliographic Record for Books,^[2] the records were divided into fields, with each field recorded on a separate colored card. The following fields were included when present on the records:

• 020 ISBN
• 050 LC call no.
• 082 DDC call no.
• 1XX Main Entry
• 240 Uniform Title
• 245 Title, other title, statement of responsibility
• 250 Edition statement
• 260 Imprint (place of publication, publisher, and date of publication)
• 300 Physical description
• 440 Series statement
• 5XX Note fields, including:
• 504 Bibliography note
• 505 Contents note
• 6XX Subject headings
• 7XX Added entries

A card environment was chosen in order to avoid any potential influence that specific user interfaces in an online environment might have on the results.

For collocation and ease of data analysis, each record was assigned a color. The Core records were orange and green, the Full records, yellow and purple. Each card set was accompanied by a questionnaire of the corresponding color. A space was provided on each card for participants to rank the usefulness of the data element. By way of example, a photocopy master for part of the Orange Core record card set is included in Appendix B. Participants ranked the usefulness of each data element according to the following scale:

1= definitely not needed
2= somewhat useful
3=no opinion
4= useful
5=absolutely essential

Participants also responded to a color-coded questionnaire for each record. The questionnaire attempted to elicit information about the sufficiency of information in each record as well as elements not included in the record that would have been useful. The first section asked participants to rank their level of agreement with six statements about each record as a whole. The second section addressed the usefulness of elements not included on the record. The questionnaire and accompanying instruction sheet can be found in Appendices C and D. Participants ranked their level of agreement with the questionnaire items according to the following scale:

1=strongly disagree
2=somewhat disagree
3=no opinion
4=somewhat agree
5=strongly agree

C. Findings

The pilot study set out to answer a number of research questions. Mean (or average) values for the data elements reflect the mean level of usefulness that users perceived for each element according to the scale given above (1=definitely not needed, 2=somewhat useful, 3=no opinion, 4=useful, 5=essential.) For the questionnaire data, mean values reflect the users' mean level of agreement for each item according to the scale shown above (1=strongly disagree, 2=somewhat disagree, 3-no opinion, 4=somewhat agree, 5=strongly agree). In the context of the pilot study, a "useful" element is defined as one with a mean value greater than or equal to 4.0 (useful to essential). It is important to note that while the answers to research questions 1 and 3 are based on a simple analysis of mean responses (Tables 1 and 3), the answers to the other questions are based on results of statistical tests rather than indicators of central tendency alone. This is because an examination of mean responses alone cannot give the complete picture where there is variability in response. A summary of the findings follows.

1. What are the most important data elements in Core and Full records as ranked by all users?

   Table 1 shows that for the Core record, the following elements were ranked by all users as most useful, with mean levels of usefulness greater than or equal to 4.0 (useful to essential):

   • Title (4.68)
   • Primary Author or Editor (4.46)
   • LC call number (4.08)

   Table 2 indicates that for Full record, the following elements were ranked as most useful by all users, with mean levels of usefulness greater than or equal to 4.0 (useful to essential):

   • Title (4.69)
   • Primary author or editor (4.57)
   • Added author or editor (4.27)
   • Series (4.08)
   • LC Call No. (4.07)
2. Did users agree on a definable set of most essential elements?

   For the most part, users in all three groups agreed upon the following set of elements as most useful across both record sets:

   • Title
   • Primary author or editor
   • Added author or editor
   • Series
   • LC Call No.

   However, the results of a one-way ANOVA and post-hoc Scheffe test revealed significant differences in the rankings of two of the essential elements according to user group. For both the Core and the Full record, undergraduates viewed the LC call number as significantly less useful than did faculty. (Table 3, mean of 3.6 for undergraduate Core vs. mean of 4.53 for faculty Core; Table 5, undergraduates Full mean 3.69 vs. Faculty, 4.35.) For the Core record, undergraduates also viewed the title as significantly less useful than did the Faculty, with a mean of 4.54 for undergraduates vs. 4.86 for Faculty (Table 3). It is difficult to interpret this finding. It could mean that undergraduates viewed the title as a less significant element in general or it could indicate dissatisfaction with the title. This finding may have less practical significance, as the mean ranking by undergraduates as a group is still well above 4.0, the value denoting a "useful" element in the context of the study.

3. How do the elements defined by the Core record standard compare to those elements ranked as most essential by users?

   The set of data elements encompassed by the Core Record Standard appears to meet users' needs quite well when compared with the set of data elements identified by users as most essential. The Core Record Standard mandates the use of all elements ranked by users as most useful.

4. Did users view either the Core or the Full level record as more useful than the other?
   • sufficient information to find the item in the library catalog (FRBR user task find),
   • sufficient information to identify the item among a group of like items in the library catalog (FRBR user task identify),
   • seeking a particular version and needing to distinguish whether or not the item is the one being sought, (FRBR user task select), or
   • sufficient author entries

   Further research is needed before definitive conclusions can be drawn about the ability of the Core record to meet users' needs in these four areas.

5. Did users find that the Core record contained sufficient
   • Author entries?
   • Title entries?
   • Subject headings?

   Results in Tables 8A and 8B appear to indicate that the information provided in the Core record is not uniformly sufficient in the eyes of users. These tables examine the sufficiency of elements in both record sets in and of themselves, in isolation from each other. Separate single-sample one-tailed t-tests were performed on the corresponding composite Core and Full questionnaire datasets to obtain this data. There is not enough evidence to support the claim that Core records contain sufficient information with regard to author, subject, and title entries from the data obtained in the pilot study. From the evidence of mean rankings alone, users would appear to be somewhat satisfied with the sufficiency of author and title entries (means of 4.1565 for author, 4.0612 for title), but somewhat less than satisfied with respect to sufficiency of subject headings (mean of 3.8912) in Core records (Table 8A). However, in the absence of further data, no definitive conclusions can be drawn with respect to user perceptions of sufficiency of these elements for Core records.

   Users do, however, appear to agree that Full records provide sufficient author, title, and subject headings (Table 8B).

6. Were there differences in response between faculty, graduate students and undergraduates with respect to:
   1. Usefulness of data elements?

      Users in all three groups generally agreed upon the usefulness of data elements in Full and Core records. However, a one-way ANOVA and post-hoc Scheffe test revealed that for the LC call number, the ISBN, the title and the subject headings, there were significant statistical differences in level of perceived usefulness across user groups (Tables 3-6). Undergraduates viewed the LC call number and ISBN as significantly less useful than did faculty for both the Core (undergraduate means 3.6 and 1.95, respectively; faculty means 4.53 and 2.75, respectively) and Full record (undergraduate means, 3.69 and 2.14, respectively; faculty means 4.35 and 2.73, respectively). For the Core record, undergraduates found the ISBN to be less useful than did graduate students (undergraduate mean, 1.95; graduate mean, 2.57). Undergraduates also viewed the title in Core records as significantly less useful than did the Faculty (undergraduate mean, 4.54; faculty mean, 4.86). This finding may have limited practical significance as the mean ranking by undergraduates as a group is still well above 4.0, the mean value defining a "useful" element for this study. The most interesting finding, however, is that undergraduates found subject headings for Core records to be significantly more useful than did either faculty or graduate students (undergraduate mean, 3.88; faculty mean, 3.52; graduate mean, 3.67).

   2. Sufficiency of information in the Core records for
   • Author entries?
   • Title entries?
   • Subject headings?

   With the exception of title entries, there was no statistically significant difference by user group in the level of perceived sufficiency. A one-way ANOVA and post-hoc Tukey tests were performed on the questionnaire data to determine if significant differences in questionnaire response occurred across user groups (Tables 9A and 9B). Statistically significant differences in response between user groups were found only with regard to undergraduates' satisfaction with the sufficiency of title entries in both Core and Full records (Tables 9A and 9B). In both cases, undergraduates were significantly less satisfied than were graduate students.

7. How well did Core and Full records support the user tasks find, identify and select as defined in the document, Functional Requirements for Bibliographic Records?

   Results of statistical tests (see Table 8A) indicate that users appear to believe that Core records contain sufficient information to allow them to find the items in a library catalog. They also indicate that Core records contain enough information for them to identify an item from among a group of like items in a catalog. Test results further show that users believe Full-level records provide sufficient information to support the same tasks (find and identify) (Table 8B). However, users did not necessarily believe that either the Core or the Full record provides sufficient information to accomplish the FRBR user task select, to distinguish a particular version of an item that is sought.

   Additionally, a one-way ANOVA and post-hoc Tukey tests indicate some significant differences in response according to user groups (Tables 9A and 9C). Undergraduates were significantly less satisfied than were faculty with the ability of both Core and Full records to provide sufficient information for identifying an item among a group of like items. For the FRBR task select, undergraduates were significantly less satisfied than faculty with regard to the ability of Core records to provide sufficient information to distinguish a particular version. They were significantly less satisfied than graduate students with the ability of the Full record to allow them to select a particular version of an item.

D. Discussion and Implications Of Findings

The next section addresses implications that the pilot study may have for the development of the Core record. It is important to bear in mind the uncertainties associated with generalizing to all users the conclusions of this relatively small study (75 users).

1. User Perception of Elements in the Core Record

   In general, the set of required elements in the Core record meets users' needs quite well. All elements that users identified as most useful (with means greater than or equal to 4.0) are required elements in the Core Record Standard. There was little difference in users' ranking of the usefulness of record elements for Core versus Full records. The Core Bibliographic Record for Books is quite similar to the Full record for books.

   However, what is much more interesting about these findings (see Tables 1 and 2) is that users ranked many data elements as less useful than would the library community. For example, users ranked elements such as imprint and edition as much less useful than expected (means of 3.81 and 3.36, respectively, for Core; 3.9 for imprint for Full), somewhere between "no opinion" and "somewhat useful." Librarians would consider both elements to be absolutely essential (5). Both record elements are clearly needed in order to distinguish works at the level of the expression or the manifestation.

   Furthermore, the pilot study showed no evidence that users believe either the Core or the Full record provides sufficient information to accomplish the FRBR user task select, i.e., to distinguish a particular version of an item that is sought. In other words, users ranked the very record elements that would allow them to accomplish this user task as less important than other elements, but they seemed at the same time to express dissatisfaction with the ability of records to support the user task select.

   These findings suggest that the users lack a clear understanding of what constitutes bibliographically distinct entities. Users appear to understand neither the functions of the various elements of bibliographic records nor how to interpret them. This suggests that users need a record structure that is easier to use and to understand than the current structure, yet one that still supports sophisticated user tasks. Further study is needed in this area.

   Results of the pilot study indicate that users perceived subject headings, with a mean of 3.69 for Core and 3.53 for Full, to be less useful than several other record elements. These means fall well below the 4.0 level that designates a "useful" element in the context of this study. There seems to be anecdotal evidence in the library community that users perform subject searches with less frequency than other OPAC searches, such as title and keyword. The pilot study's finding is consistent with that observation. The reason for the relatively low means for subject headings in both record sets could be that users tend to rely more heavily on known item searching than on subject searching.^[3]

   Not all user groups viewed the usefulness of subject headings similarly. Undergraduates found them to be significantly more useful for Core records than did faculty. Faculty and graduate students are more likely to perform known item searching, whereas undergraduates are less familiar with subject areas and may do more subject searching than those in the other groups.

   Undergraduates were also significantly less satisfied with title access for Core records than were faculty. This finding is difficult to interpret. On a practical level, it may have limited significance, as undergraduates clearly still perceived the title as useful to essential, with a mean of 4.53. Several interpretations are possible. Perhaps undergraduates found the title as an element or data category to be less useful than did faculty. Possibly, they were simply dissatisfied with the sufficiency of title access on the Core record. Consistent with the earlier finding regarding the usefulness of subject headings for undergraduates is the possibility that they find title searches less useful. Further research is needed here.

   Of further interest is the fact that with respect to the sufficiency of title access users were satisfied with Full records but not necessarily with Core records. Contents notes were included on Full records but not Core in the pilot study. The Full record contents note had a mean of 3.87, which is close to mean value 4.0 that identifies a "useful" element. It is possible that users were happier with the sufficiency of title access on the Full record because it included a contents note. One final observation concerning titles is that users did not consider either series or uniform titles to be particularly useful elements (only on the Full record was series ranked as useful). Again this suggests that users may not have a clear understanding of the functions of title elements.

   Undergraduates found the LC call number on Core and Full records significantly less useful than did faculty, with means of 3.6 for Core and 3.69 for Full. The other users considered the LC call number one of the most useful elements. Perhaps undergraduates do not understand the relationship of the LC call number in the catalog record to the item's shelf location as well as do graduate students and faculty.

2. User Satisfaction with Core Records Overall

   Users seemed quite satisfied with the set of data elements included in the Core Record for Books. Users were also as satisfied with the Core record overall as with the Full except with regard to title and subject access. Results of the pilot study indicate that users found the Full record to be more useful than the Core record with regard to the sufficiency of both title entries and subject headings (Tables 10, 8A, 8B). Statistical analysis of users' perceptions of the Full record in and of itself showed them to be satisfied with the sufficiency of author, title, and subject entries. However, the same could not be said with statistical certainty about their perceptions of the sufficiency of author, title and subject headings in the Core record (Table 8B).

   For monographs, the Core and the Full records are quite similar. The main areas of difference include the subject headings, added entries and notes typically used for each record type. For the Core record, the decision of whether to add a note, an added entry, or a subject heading is very much dependent upon the cataloger's judgement. In particular, the area of subject analysis in the Core has been somewhat controversial. David Banush's recent study on practitioner perspective of the Core revealed that misunderstandings about the requirements for subject headings in the Core persist.^[4] The Core requires that every record must have at least one or two subject headings. The Core does not limit the number of subject headings permitted on a record: this is a minimum, not a maximum, requirement. Catalogers and cataloging agencies may add subject headings to Core records as they judge necessary.

   The results of the pilot study indicate that users are satisfied with the sufficiency of subject headings on Full records in and of themselves (Table 8B). Users found Full records to be more useful than Core records with regard to sufficiency of subject headings (Table 10). But as to user satisfaction with the sufficiency of subject headings in the Core in and of itself, no conclusions can be drawn with statistical certainty from this study (Table 8A). It appears that users might prefer to see more subject headings in Core records. However, further research in this area is needed before definitive conclusions can be drawn.

   Additionally, users indicated a preference for Full records over Core with respect to the sufficiency of title entries (Table 10). This finding is difficult to interpret, since users did not seem to find series titles (for Core records) or uniform titles particularly useful (Table 3, Table 5). Users seemed reasonably positive with regard to the usefulness of contents notes (mean of 3.87), which are not required for Core records for single volume works. The use of contents notes in Core records could be a means of improving title access to these records. This is an area where further research is needed. Overall, users were very satisfied with the Full record, but somewhat less satisfied with the Core.

3. Does One Size Fit All? User Groups and the Core

   The pilot study revealed statistically significant differences in response between user groups in a number of areas (Tables 4, 6, 9B, 9D). Both undergraduates and faculty seemed to have special needs for catalog records. Undergraduates were much less satisfied with records of either type than were graduate students or faculty. They seemed to perceive record elements (e.g., the title and the LC call number) as less useful than other users. However, subject headings were significantly more useful to undergraduates than to other users.

   The differences between user groups identified by the pilot study raises the question of how well Core records are able to meet the needs of disparate user groups. Enhancements to Core records may be needed for certain collections, types of materials or subject areas in order more effectively to meet the needs of various user groups. More research is needed in this area.

4. Record Quality and the Core Record

   A key question about the Core Record for practitioners and users alike is what constitutes a quality bibliographic record. David Banush's study attempted to explore this question from the perspective of catalogers and managers.^[5] The pilot study has attempted to address this question from the user's perspective, through consideration of a number of key issues. The study has examined the constitution of data elements in the Core: does the Core include the elements users identify as most useful? It has examined the question of sufficiency of data: Does the Core provide enough information to meet the needs of users? The study has also tried to assess the functionality of the Core: how well does it support common user tasks such as find, identify, and select? The pilot study has provided a number of insights into the definition of "quality" bibliographic records from the user's perspective. Data from this study suggest that a quality record from the user's perspective includes a comprehensive set of data elements, a high level of functionality with respect to the selecting task, and a simpler structure. However, further research is needed in order to arrive at a more conclusive answer to the question of record quality from the user's point of view.

5. Research Methodology

   The researchers feel that the methodology used in the pilot study was fairly successful, yielding useful data. More work is needed in refining the methodology. One area that should be revised is the scales used to rank the data in the card sets and questionnaires. The scales were designed in consultation with a faculty member of the SMSU Department of Psychology. However, the researchers found that placing the value "no opinion" in the middle of the scales made it difficult to interpret responses. "No opinion" should instead be placed outside the scales.

   Another modification suggested by members of the PCC Standards Committee at the ALA Annual meeting in San Francisco, June 2001 would be to show users an entire record rather than to use a separate card for each data element. This is an option that could be pursued if the study were to be expanded.

   The questionnaire yielded the most useful data obtained in the pilot study, as it addressed the information in Core and Full records as a whole. The question of sufficiency of data--what is "enough" to meet the needs of users?--is central to understanding the impact of the Core on users. It is also quite difficult to design a research methodology that addresses the user tasks effectively, the chief difficulty being that these user tasks are not actually performed discretely in a real-life situation. The study could be improved through further development of the questionnaire, particularly with respect to the user tasks.

RECOMMENDATIONS

The pilot study is based upon the perceptions of 75 catalog users who examined four catalog records, an admittedly small sample. The recommendations below are offered for the committee's consideration based upon the available data. While users responded quite positively to the Core, the pilot study highlighted some areas of concern.

1. As is the case with much research, the pilot study raised more questions than it answered. Further research is needed on the impact of the Core record on users. The pilot study could be expanded to a larger and more diverse group of users, from a variety of geographic locations and sizes of institution, in order to obtain further data. Research is needed in a number of areas, including:
   • ability of Core records to support user tasks
   • sufficiency of Core records with respect to author, title, and subject entries
   • user understanding of the functions of record elements
   • ability of the Core to meet the needs of various user populations
   • benefit of authority work in Core records to users

   Further work is needed to determine what constitutes a quality bibliographic record from the user's perspective, and to assess how well the Core record meets that definition. In many areas, the data obtained from the pilot study were not sufficient to answer these questions conclusively; however, the study contained useful methodology that could be refined and expanded.

   Two key areas where further research is needed are highlighted in recommendations 2 and 3. These recommendations address the areas of the Core where users were less satisfied with Core records than with Full. Recommendations 2 and 3 are offered on the basis of data from the questionnaires, specifically, findings in Tables 8A, 8B, and 10. The pilot study found that users were as satisfied with Core records as with Full with the exception of the sufficiency of subject and title entries. In Table 10, user satisfaction with the sufficiency of information in Core records was compared directly with that in Full. The data showed that users clearly preferred Full records to Core with regard to the sufficiency of subject and title entries. In other words, they found that Core records were less useful than Full with regard to the sufficiency of these elements.

   Tables 8A and 8B examine the sufficiency of data elements in Core and Full records in and of themselves, considered in isolation from one another. Table 8B shows that users felt Full records provide sufficient author and title entries. Table 8A shows that there are not sufficient data to support the claim that users found the same to be true for Core records. It may be the case that they considered Core records sufficient in these areas, but the available data do not support this conclusion. More research is needed here.

2. The area of sufficiency of title access to Core records needs further research. A specific question to consider is whether it is advisable to expand the Core Record for Monographs to include contents notes for single volume works. Tables 8A and 8B show that users were satisfied with the sufficiency of title entries on Full records but the data were inconclusive for Core records. The main difference in title access between the Core and Full records in the study was the contents notes. Contents notes (with mean ranking of 3.87) were ranked close to the mean value 4.0 that identifies a "useful" element (Table 2). Currently, contents notes are required in the Core only for multipart items with separate titles. The addition of contents notes to the Core record for single volume works as well might improve user satisfaction with the sufficiency of title access to Core records.
3. The sufficiency of subject access to Core records from the user's perspective is another area where further research is needed. The pilot study indicated that users found Full records more useful than Core with respect to the sufficiency of subject headings (Tables 8B, 10). The study also showed that subject headings for Core records were significantly more important to undergraduate users than to the other user groups (Tables 3 and 4). These findings must be weighed against the fact that subject headings were not ranked by users as a "useful" element in the context of the study, since the mean levels for both record types fell below 4.0 (see Tables 1 and 2). However, users clearly preferred Full records to Core with regard to subject access (Table 10). This seems to suggest that users might prefer to see more subject headings on Core records since they appear to believe that Full records contained "enough" subject headings (Table 8B). Improving subject access to Core records may address the needs of a greater number of users.

   Furthermore, there is a widespread misconception in the cataloging community that the Core limits the number of subject headings that can be included on a record. Catalogers and cataloging agencies need to be made more aware that the Core is not a ceiling, but a floor, and that core records can be enhanced to meet local needs. The idea that Core records are dynamic needs to be emphasized.

   The researchers offer two suggestions for improving subject access to Core records.

   1. Increase the awareness of the cataloging community of the Core as a dynamic record type. Encourage cataloging agencies to consider whether the development of local policies for the enhancement of Core records in the area of subject analysis is appropriate for their user groups.
   2. Clarify the wording of footnote 10 in the Core Bibliographic Record for Books (http://www.loc.gov/catdir/pcc/bibco/corebook.html), which states:

      "If appropriate, assign at least one or two subject headings from an established thesaurus or subject heading system recognized by the MARC 21 Format."

      Consider revising the footnote to read:

      "If appropriate, assign at least two subject headings from an established thesaurus or subject heading system recognized by the MARC 21 Format."

Revised and submitted 10/23/01