Test of the BIBCO Standard Record

June 14, 2010

A study conducted by:  Diana M. Brooking (project leader), Joseph A. Kiegel, Adam L. Schiff

Introduction

The BIBCO Standard Record (BSR) for textual monographs was implemented as a cataloging standard by the Program for Cooperative Cataloging in January, 2010.  It replaced the former Full and Core standards, and is meant as a floor record that provides essential data elements to support the FRBR user tasks of finding, selecting, identifying and obtaining resources.  A floor record means that catalogers may (but are not required to) add additional elements  above the standard based on local policy or cataloger's judgment.  The PCC Policy Committee expects to approve BSRs for additional formats at the 2010 ALA Annual Conference.

Unlike the CONSER Standard Record, which underwent a pilot project before its implementation (http://www.loc.gov/acq/conser/alrFinalReport.html), the BSR was not tested prior to adoption.  Since the BSR is quite similar to the Core standard, and there are differing opinions whether Core saves much time in comparison to the Full record, we decided to conduct a test of the BSR on the scale of a single institution.  The results of the test were intended to guide us in making a decision about adopting the BSR locally.

Method

Four experienced original catalogers were recruited, given a training session on the BSR, and asked to practice creating records at the BSR level and to practice timing themselves on full records.   

A set of 12 test titles was developed.  They were in English, on relatively general subjects in the social sciences and the natural sciences, and included examples of various cataloging situations such as series and conference headings.  None of the titles had cataloging copy in OCLC.  They were drawn from PDF documents on the Web, but were given to catalogers as printouts and cataloged as printed books rather than as Web resources.  Test records in foreign languages were deliberately omitted because we do not have enough catalogers with sufficient language expertise to sample across a range of catalogers.

Local policy decisions were made for the purpose of the test that implemented the BSR as a true floor, in order to provide a fair test and highlight the differences between Full and BSR.  For example, we did not provide added entries for government bodies when the only function was as publisher.  We did not provide added entries for related works.  For subject headings, we limited them to primary topics and did not use the "20% rule".  We did, however, follow the rules of LCSH, e.g. giving a full array of subject headings called for (paired headings, etc.), and  AACR2/LCRIs for guidance on "when deemed important to users," e.g. for title added entries.

The twelve test titles were divided into four batches of three printouts each.  Every cataloger was assigned to catalog a given batch at either Full or BSR (two batches at Full and two at BSR per cataloger) in such a way that each test title was done twice at each level (see chart below).  This variation was designed to provide some compensation for natural differences in cataloger speed and subject expertise.

Catalogers completed their work on the test records, keeping track of the time spent on cataloging and authority work to the nearest minute.  It was important to include authority work because this may vary between Full and BSR due to choices in the headings used.  Catalogers worked in OCLC Connexion and saved their bibliographic and authority records in a local save file, so that they were not able to view each other's work.  After the cataloging was completed, the number of authority records created and the number of subject and name headings assigned were counted for the Full and BSR records.

Results were analyzed using PASW Statistics 18, by SPSS, Inc.  The frequency distributions of the values of the variables Time, Authorities, Subjects and Names were examined and found to be not normal.  Therefore, a standard T-Test could not be used to compare the means of the cataloging levels (Full vs. BSR).  Instead, a nonparametric test was used, the Indepedent Samples Mann-Whitney U Test, which was chosen by the software and does not require an assumption of a normal distribution.  

Results

Quantitative Results

Time

For each cataloger, we calculated the average time spent on their six Full records and their six BSR records, and then found the percent reduction in time between the two levels.  The reduction in time ranged from 11% to 49%, with an overall average of 26% less.  We tested the difference between the mean times for Full vs BSR and found it to be statistically significant (p = .021).

Subject Headings

For each cataloger, we found the average number of subject headings applied to Full and to BSR records, and calculated the percent reduction, which ranged from 0% to 47%, with an overall average of 30% less.  The difference between the mean number of subject headings for Full vs BSR is statistically significant (p = .019).

Name Headings

In similar fashion, we found the average number of name headings applied (1XX and 7XX fields).  The reduction in name headings ranged fron 17% to 34%, with an overall average of 24% less.  The difference between the mean number of name headings for Full vs BSR is not statistically significant (p = .110).

Authority Records 

We also found the average number of authority records created for Full and BSR bibliographic records.  The range of difference was from 100% increase for the BSR to a 62% reduction, with an overall average reduction of 17%.   The difference between the mean number of authority records for Full vs BSR is not statistically significant (p = .522).

Qualitative Results

The quality of the records created in the test was high.  Only one error was discovered during a post-test review, where a cataloger overlooked a series on the piece and did not enter it.  There was some variation in how the BSR was applied, which was found chiefly in the notes area.

The reviewers were struck by the degree of variance in the records beyond differences in the Full and BSR standards.  Catalogers have options under the rules and room for judgment, and legitimate variations were common.  For example, the title "Clusters--a key to rural prosperity" was recorded three ways in field 245 by different catalogers:

Clusters, a key to rural prosperity

Clusters : $b a key to rural prosperity

Clusters--a key to rural prosperity

Subject analysis, which has a greater impact on cataloger effort, showed variation.  On some records, it was quite consistent across levels.  For example one title had two subject headings on both BSR records:

Income $z Washington (State) $v Statistics

Poverty $z Washington (State) $v Statistics

while the Full records had these headings, and one of the records had an additional heading:

Washington (State) $x Economic conditions $y 20th century

In other cases, the core subject idea was identified consistently, but there was variation in emphasis and how it was expressed across the records:

BSR 1:

International business enterprises $v Congresses

BSR 2:

International trade $v Congresses

International business enterprises $v Congresses

Full 1:

International business enterprises $v Congresses

Personnel management $v Congresses

Conglomerate corporations $v Congresses

Full 2:

International trade $v Congresses

Globalization $v Congresses

International business enterprises $v Congresses

Discussion 

The BSR clearly saved time and effort in comparison to the Full record.  The average savings in time across the catalogers and test records was 26%, or a full quarter of the time, and the result is statistically significant.  The number of subject headings, which averaged 30% less and with a difference that  is statistically significant, clearly contribute to the savings.  Beyond that, with the test design used, we cannot attribute the relative percent of savings to reductions in description, in name headings or to authority work.  They all played some role, but we are not able to say what had the most impact and what the least.

Name headings assigned and authority records created also averaged lower on BSR records, by 24% and 17%, respectively.  However, neither difference was statistically significant.  For name headings, this may be because catalogers have less leeway in the number of access points they assign, that is, there is a maximum number dictated by the piece itself (the names appearing on it), whereas subject headings are not limited in the same way.   For authority records, once a choice of name headings is made, the cataloger has no control of the number of authorities created.  While there was some decrease on average as expected, since there are fewer name headings on BSR records, individual catalogers saw a wide variation, including twice as many authorities for BSR records as Full for one cataloger.  Thus a reduction in authority work does not seem to be a necessary corollary of using the BSR, and is more dependent perhaps on the subject or language involved.

There was some variation in how the BSR was applied in the test, in particular in the notes area.  This may have been due to the conditions of the test, in that catalogers did not have equivalent experience using the BSR as they had with the Full standard.  We expect, however, that even with increased experience there would be variation in application of the BSR among any group of catalogers.

Finally, the natural variation in practice, due to legitimate cataloging options and judgment, is substantial.  It confounds the notion that there is "one" Full record and "one" BSR record for a given title, and a fixed difference between the levels. 

Conclusion

Overall, the BSR saves time and may be employed to reduce cataloging effort.   However, its implementation should be tempered by an understanding that there is a range of cataloging practice.  Since this test was conducted on the scale of a single institution, its results are not necessarily generalizable to other institutions, subject areas, languages, or formats.  We believe testing should be conducted at other libraries to get a better sense of the benefits and limits of the BSR.  We found the testing was not labor-intensive and think others could replicate it without difficulty.

Appendix

Practice Titles:

Dyslexia: A Discussion and Resource Paper
http://www.ode.state.or.us/pubs/sped/dyslexia.pdf

KOI 2006: 11th International Conference on Operational Research
http://oliver.efpu.hr/koi06/koi06_proceedings.pdf

Landslide Hazards in the Stillaguamish Basin: A New Set of GIS Tools
http://www.stillaguamish.nsn.us/Publish/landslide%20hazards.pdf

Test Record Set:

Changes in Income and Poverty in Washington State 1989-1999
http://www.crs.wsu.edu/wacts21/EB1973E.pdf

Clusters--A Key to Rural Prosperity
http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/all/csi12897/$FILE/Clusters-%20A-key-to-Rural-Prosperity.pdf

Coastal Protection and Development Planning Guidelines for Bermuda
http://www.planning.gov.bm/documents/Guidance%20Notes/DC%20Coastal%20Development%20Protection.pdf

Forced Out: LGBT People in Armenia
http://ilga-europe.org/home/publications/reports_and_other_materials/forced_out_lgbt_people_in_armenia_february_2009

Freshwater Mussels of the Pacific Northwest
http://www.xerces.org/wp-content/uploads/2009/06/pnw_mussel_guide_2nd_edition.pdf

Historic Town of St. George and Related Fortifications: Management Plan
http://www.planning.gov.bm/documents/Historic%20Environment/world_heritage_site_management_plan.pdf

Internet Voting in the March 2007 Parliamentary Elections in Estonia
http://www.vvk.ee/public/dok/Coe_and_NEC_Report_E-voting_2007.pdf

Proceedings of the 14th Australasian Vertebrate Pest Conference, Darwin,
Australia
http://www.invasiveanimals.com/downloads/Final-proceedings-with-cover.pdf

Proceedings of the 48th Annual Meeting of the Academy of International
Business
http://aib.msu.edu/events/2006/AIB2006_Proceedings.pdf

The Rights of Children Raised in Lesbian, Gay, Bisexual, or Transgender
Families: A European Perspective
http://www.ilga-europe.org/home/publications/reports_and_other_materials/the_rights_of_children_raised_in_lesbian_gay_bisexual_or_transgender_families_a_european_perspective_december_2008

Swedish National Seismic Network (SNSN): A Short Report on Recorded
Earthquakes during the Fourth Quarter of the Year 2009
http://www.skb.se/upload/publications/pdf/P-10-04webb.pdf

Transportation Acronym Guide
http://www.mdt.mt.gov/publications/docs/brochures/tag.pdf