Prepared by the Core Cataloging Task Group (CCTG)December 16, 1996
The experiment itself. For the six-month period May- October, 1996, thirty staff members participated in the Core Level Cataloging Experiment. 1548 records were completed with an expenditure of 2448 hours for a productivity of .63 titles per hour (data from STARS). This compares favorably with a directorate-wide productivity of .36 for full level (original) cataloging and .88 for copy cataloging. Material cataloged during the experiment was primarily limited to items slated or eligible for minimal level cataloging (MLC) and included a broad range of languages and classes of books (exclusive of belles lettres), music scores, and sound recordings. Assuming that the application of core level equates to a savings, the results of tabulating information from 1541 STARS slips (seven slips were lost during processing), shows that core level cataloging represents some kind of savings in the descriptive aspect 51% of the time (793 items) and in the subject aspect 18% (276 items) of the time.
Staff applied the national core level standard augmented by additional data elements. This augmented standard is referred to as the "LC core level" or an "LC core level record."
Results. The results of the experiment can be characterized overall as mixed and undramatic but with enough positive evidence to indicate that use of core level cataloging has potential worth pursuing. The productivity results are quite encouraging but are to be tempered with certain caveats. The task group also used another means, a statistical program called JLGSTATS, to attempt to determine if core records represented a bibliographic savings. Although the data analysis of these runs produces no striking findings to explain the productivity results, it nevertheless reveals solid, concrete evidence that the cumulative effect of core level can indeed impact productivity. Team responses are mixed, with more positive than negative reactions, but with a certain uneasiness over the introduction of yet another ingredient in the already overly complicated recipe for the bibliographic cake. Because the material for the experiment was pre-selected, no time was spent determining which cataloging mode to apply. Thus the experiment did not show the time it will take to determine cataloging mode on an item-by-item basis for materials with a mix of priorities. There is consensus among the teams that the pre-selected materials did not reflect current, day-to-day receipts. Therefore, it was very difficult to extrapolate the core experience to the wider universe with any confidence.
Recommendations. The Core Cataloging Task Group (CCTG) recommends:
This Final Report from the Core Cataloging Task Group (CCTG) records the results of the Core Level Cataloging Experiment designed by the CCTG and described in its Report to the Acting Director for Cataloging dated March 11, 1996. The Final Report also includes recommendations from the CCTG to the Cataloging Management Team (CMT) based on the results of the experiment itself.
Description of the experiment. For the six-month period May-October, 1996, thirty staff members participated in the Core Level Cataloging Experiment. 1548 records were completed with an expenditure of 2448 hours for a productivity of .63 titles per hour (data from STARS). This compares favorably with a directorate-wide productivity of .36 for full level (original) cataloging and .88 for copy cataloging. Material cataloged during the experiment was primarily limited to items slated or eligible for minimal level cataloging (MLC) and included a broad range of languages (see Appendix 2.0, p. 2 for the languages included) and classes of books (excluding belles lettres), music scores, and sound recordings.
Standards applied during the experiment. Participants applied the core level standards developed for books in roman script, books in nonroman scripts, music scores, and sound recordings. These standards were developed by various Task Groups of the now PCC Standing Committee on Standards. (The task group assigned to develop the JACKPHY standard was unable to reach a consensus regarding romanization. Some members thought that it was unnecessary to romanize other title information and statements of responsibility if they were present in a non-roman form. LC members of the task group, on the other hand, urged complete romanization of all data present. For the LC core experiment, staff included and romanized all data occurring in other title information and statements of responsibility.) For the LC core experiment, staff applied, as appropriate, additional elements (050 alternate number; 504; in 6XX fields, assigned paired headings to express reciprocity between two places; 082; 041; 043; FFD 02 (Form of composition); FFD 03 (Format of music); 024 0 (International Standard Recording Code), if present on item)). This augmented standard is referred to as the "LC core level" or an "LC core level record."
Participants in the experiment. Staff from the following teams and divisions participated in the experiment:
Physical Science Team I and Technology Team I, ASCD
Romance Team, HLCD
Hebraica Team, RCCD
Central & Eastern European Languages Team, SSCD
Germanic & Scandanavian Languages Team, SSCD
Romance Languages Team, SSCD
Music & Sound Recordings Teams I and II, SMCD
STARS statistics. During the experiment, statistics were recorded twice: once in the regular STARS system and again in a separate STARS system used specifically to keep core level cataloging statistics. A special STARS slip was used in conjunction with the core level STARS system. Participants were asked to check a table indicating whether the descriptive and subject aspects of each item done core level was the same as full level or not.
Assuming that the application of core level equates to a savings, the results of tabulating information from 1541 STARS slips (seven slips were lost during processing), shows that core level cataloging represents some kind of savings in the descriptive aspect 51% of the time (793 items) and in the subject aspect 18% (276 items) of the time. For additional statistical details taken from the core STARS system, see Appendix 1.
JLGSTATS. James L. Godwin wrote and, over the years, has maintained a statistical program called JLGSTATS. In general, the program provides two kinds of counts:
Ardith Bausenbach, APLO, at the request of the CCTG, did several runs of JLGSTATS against various datasets to provide detailed statistical data element counts of materials in the core experiment. Since the JLGSTATS are run against all the files that constitute the current online catalog once a year in October, statistics from those runs were also used in the comparisons. In addition, Ms. Bausenbach did special runs that made it possible to show and compare the various counts across different categories of records, e.g., all those in a particular file, all those of a particular type, categories of full level and core level. The detailed results of these runs are given in Appendix 2 (Summary Data; Headings (6XX/7XX)). A description of the data elements counted (given in the rows of the tables) and the categories of records (given in the columns) in the tables for summary counts and heading counts appears in Appendix 2.0; an analysis of these counts organized by the data element counted and subarranged by the category appears in Appendix 2.9. Tables of counts for selected variable fields were also compiled from the JLGSTATS runs but have not been included in the Final Report because of their extent. They can be made available to anyone who wishes to see them. The data in the following analysis relating to the 500 fields are taken from these tables.
Note that the total JLGSTATS counts for core records in various categories do not equate to those same counts from STARS. The discrepancy occurs because the counts stated in STARS are based on what staff think they did while the JLGSTATS counts reflect what they actually did. For example, STARS may indicate that a record was completed. In actual fact it may indeed have been completed but the record was not verified. Such a record would be excluded from the JLGSTATS counts, since those counts were only against verified records.
General. With a few exceptions, the JLGSTATS are, in general, concrete confirmation of aspects of core level cataloging that heretofore have been suggested intuitively. The following analysis consists of comparisons between various related categories and shows that, for the most part, as one would expect, the figures for non-core categories are greater for almost all of the data elements counted than are those for core level cataloging. In some cases, the differences are so close that they do not bear very much significance. In the following analysis, those areas in which the data show a result different from what one might expect are indicated by a vertical bar in the left margin. See Appendix 2.0 for a description of the categories being compared and for the data themselves.
All BOOKS / All Roman Bks Core.
All BOOKS is less than All Roman Bks Core in the following categories:
Comment: That core level is larger than All BOOKS in these counts is probably explained by the fact that 1) belles lettres were excluded from the core experiment and 2) the impact of the presence in the BOOKS file of 435,231 minimal level records (as of October 1, 1996).
All BOOKS is more than All Roman Bks Core in number of different tags (155 / 48).
Comment: The dramatic difference between All BOOKS and core in the overall number of tags occurring in the two categories confirms the more limited approach that characterizes core level.
All BOOKS is more than All Roman Bks Core in average number of 500 fields per record (0.65 / 0.21).
Comment: The difference between All BOOKS and core again confirms the more limited approach of core.
Full Roman Bks / Core Roman Bks (100 record comparison in German H, Italian D, and Russian Q).
Full Books is more than Core Books in the following categories with exceptions noted:
Comment: Fairly clear evidence of the more limited use of notes in core level; also illustrates variations related to different kinds of materials.
Comment: Evidence of a more limited subject approach under core.
Comment: Mixed evidence of a more limited descriptive access approach under core.
Comment: Again, mixed evidence of a more limited approach under core.
All Hebrew Full Bks / All Hebrew Core Bks.
All Hebrew Full is more than All Hebrew Core in the following categories:
Comment: Except for average 6XX per record, these counts tend to confirm the more limited approach of core.
Full Hebrew Bks / Core Hebrew Bks (class B 65 record comparison; class H 35 record comparison).
Full Hebrew Books is more than Core Hebrew Books in the following categories with exceptions noted:
Comment: No explanation for some of the counts that are contrary to what one might expect, particularly in the class H records, comes readily to mind other than that the sample is simply too small to be of much significance.
All Scores Full / All Scores Core.
Comment: The comparisons between full and core for scores and sound recordings reflect what one would expect in all the counts, i.e., all those for full are larger than those for core.
All Scores Full is more than All Scores Core in the following categories:
Full Scores is more than Core Scores (100 record comparison) in the following categories:
All Sd Recs Full / All Sd Recs Core.
All Sd Recs Full is more than All Sd Recs Core in the following categories:
Full Sd Recs is more than Core Sd Recs (50 record comparison) in the following categories:
One component of the experiment was to have participants, either individually or as a team or both, submit any observations or recommendations they cared to make related to their experience with the experiment. Responses from the participants varied, due in part to the range of materials that were included in the experiment. A summary of the responses is provided in Appendix 3.
Another component of the experiment was to explore working with reference/selection staff to alert them to the experiment itself, and particularly, to work with them in determining mutual agreement on the materials that would be appropriate to core treatment.
The Hebraica Team was able to take advantage of its already close ties with its counterpart custodial division in orienting staff in that division to the core experiment and discussing materials that would be appropriate to that treatment.
The SSCD Romance Languages Team team leader meets periodically with staff from the Hispanic Division, but was not able to undertake any special activity with respect to the core experiment.
The Physical Sciences I Team already had an on-going relationship with science reference specialists, who have helped deselect Slavic materials in the past. One of their staff attended the Core Experiment briefing the team leader did for the team and the team leader spoke about the experiment a few weeks later to a meeting of Science Reading Room staff and ASCD team leaders. The reference staff were receptive to core because it was applied to a large Slavic arrearage and there would otherwise have been no records at all if it were not for core.Go to Top
Although the productivity rate and perceived savings for the experiment are encouraging, the CCTG offers the following observations in interpreting the results.
Participants. Most of the participants on the various teams were volunteers (and perhaps highly motivated) and were of a very high calibre, so that higher productivity is not necessarily or strictly related to the fact that core level guidelines were being applied. It is not unreasonable to expect that these same staff might well have a very similar level of productivity doing original, full level cataloging.
Recording of time. Extensive efforts were made to keep the recording of time consistent with the way "regular time" is recorded, including time spent on non-production work. Nonetheless, since the recording of STARS hours by and large was limited to a single activity, i.e., cataloging, the productivity may not have included the usual non-productive aspects of cataloging, e.g., informal consultations, documentation update, training and review, cyber-activities, etc. Therefore, the productivity rate may be artificially high.
Index. Although not specifically recorded, in a large number of instances the only DESCRIPTIVE difference between full level and core level was the absence of an index note in the latter. Given that
the descriptive "savings" must be looked at with skepticism, since anecdotal evidence suggests that for a significant number of records the only difference between full and core level was the absence of an index note.
Pre-determined material. For the experiment, the material to be cataloged at the core level was, for the most part, pre-determined. Were core level to be applied to the entire universe of material, the added factor of deciding what material should receive core level will be time-consuming to some degree, an impact that was not relevant to, and therefore not measured by, the experiment.
An additional set of guidelines. An additional set of guidelines combined with several different cataloging streams may be perceived by staff as burdensome and counter productive to any efficiencies that otherwise might be realized.
Basis for applying core and other levels of cataloging. The book materials selected for the experiment were limited primarily to items slated or eligible for MLC. While the score material was limited to the Music Teams' priority 4s, the sound recording material covered the Music Teams' priority 3s and PLC items. Therefore, it is felt that the experiment itself did not realistically test the application of core level cataloging to all materials as described in Cataloging Priorities dated January 23, 1996 and issued by the Acting Director for Cataloging under a covering memorandum dated January 24, 1996. The CCTG's interpretation of that document results in the following:
While the recommendations in Cataloging Priorities result from an analysis of book materials, these recommendations do not always apply to music and sound recordings.
Recommendation: The CCTG recommends that the CMT affirm (or clarify) this interpretation of the application of cataloging modes. If its interpretation is correct, the CCTG assumes the following impact:
Recommendation: The CCTG further recommends that core level be made available for use at the discretion of the team/division. The CCTG also recommends that after the core level is introduced across the Cataloging Directorate and teams have gained actual experience, that they be asked, working with division management and appropriate reference/selection staff, to develop more formal guidelines to refine the basis on which the team applies core level.
"Mixed bag" application of core level. Experience suggests, and the experiment confirms, that application of core level in many instances will result in a "mixed bag" application with respect to the descriptive and subject aspects. This means that there will be many cases in which the record formally labelled "core" reflects core level in the descriptive aspect but not in the subject aspect. By the same token, we currently have cases in which records formally labelled "full" reflect full level in the descriptive aspect but not in the subject aspect (e.g., records from overseas offices in which the subjects are assigned by someone who does not know the language on the basis of a brief summary created by someone who does know the language).
Recommendation: To minimize the possible impact caused by an undue concern over characterizing so-called "mixed bag" records, the CCTG recommends that we follow the pattern already established with overseas materials and base the aspect of "coreness" primarily on the descriptive aspect. Another reason for recommending this approach is that the descriptive aspect is more readily observable as perhaps not being core than is the subject aspect. This issue appears to be most relevant to those teams in which the subject aspect is often done before the descriptive aspect, and it seems prudent to minimize the number of questions that could be raised about "why is this record labelled core"?
"Fluid/static" application of core level. Results of the experiment suggest that there are two views of the application of core. One view holds that, although some flexibility is provided in assigning headings (both descriptive and subject) core level is essentially a prescriptive set of guidelines beyond which one does not go (e.g., the only notes permitted are those specifically stated). This can be called the "static" view. Another view sees core as a basic guideline to be followed in most cases but with discretion occasionally to supply a data element beyond that provided for in the guidelines. This can be called the "fluid" view. For example, although notes are limited to specific categories, there may, on occasion, be an instance in which another kind of note is judged important enough to be included in a particular record. Recommendation: Were this interpretation to be adopted, the CCTG recommends adding a further guideline at the end of the DCM stating:
After having first applied the above guidelines, one or two additional data elements may be assigned provided that failure to do so would omit extremely important information about the item.
This may result in a further uneven meaning of core, particularly in relation to its application by other libraries, but that is already occurring with the concept of an "LC core level record."
Authority records for uniform titles. The CCTG assumes that the application of core level does not change the current limitations in making authority records for uniform titles.
Recommendation: The CCTG recommends that the CMT confirm this assumption. A change in current practice would substantially impact the application of core level, particularly to records for music and sound recordings.
Training. The core level experiment was conducted by volunteers with a minimum of training. This may have consequences for the introduction of core level cataloging across the Cataloging Directorate and suggests to the CCTG that careful planning will be required to introduce the full staff to the concept of core and the training time allotted may be more than that required for the experiment. Recommendation: Where possible, the CCTG recommends that participants in the experiment be involved in the training to provide other staff the benefit of their experience.
External core imported for use in LC. The CCTG assumes that with the establishment of an "LC core level," core records imported for use at LC will be reviewed to determine 1) the acceptability of applying core level to the item in hand and 2) if core is determined appropriate, it meets the LC core level.
Core statistics keeping (hours). During the core experiment each team used a dedicated STARS computer to keep core statistics. If, however, core cataloging were to be implemented at LC across the board, STARS would be of limited use for keeping core statistics. As currently configured, STARS could be used to provide only the number of LC records cataloged as original core. Neither the number of core records processed using copy could be provided nor could productivity be generated.
JANUS runs could provide the number of LC records cataloged as original core. They could also provide the number of core records processed using copy, distinguishing how many were PCC and how many were non-PCC. These counts will be relatively accurate, depending upon the time frame involved (e.g., whether fiscal year or calendar year). These data are at the directorate level only, not at the division and team level. Productivity could not be generated.
An ILS will presumably be able to provide statistical data but at what level and in what detail are not yet known.
Use of core level at LC. Core level cataloging should be implemented at LC. The results of the core experiment show that while the core level standard may not lend itself to straightforward application, some catalogers feel it can yield savings, with certain caveats noted. The national core standard does provide for a basic level of cataloging, but it falls to the cataloger to judge how to apply the core level adopted by LC. In some ways, this situation is not unlike full level cataloging in that different catalogers, based on experience and judgement, may come to different conclusions regarding the importance of notes, added entries, subject analysis, and the need for subject/class/development. While results from the experiment tend to confirm that the limitations on certain notes and added entries provide for savings in descriptive cataloging but that the savings in subject cataloging may not be as substantial, the overall question may be how much is gained and how much lost in adding another category for teams to manage and to catalog under.
In general, teams and catalogers participating in the experiment appear willing to consider core level for at least some of their work, providing they have the option to add to the cataloging record data elements they deem necessary for any particular item. While some teams may be able readily to identify certain types of materials that lend themselves to core treatment without essential loss in access, other teams may not. For most material, all teams will need to judge the applicability of core on a case-by-case basis and decide whether to provide basic or augmented core. Another result could be a mixed core record, with the descriptive part done as core and the subject part done as full or vice versa. This is not necessarily a bad thing, although the encoding level will identify it as a "pure" core record.
Selection of material. Management should give teams a large measure of autonomy, tempered by consultation with reference/selection staff, in deciding not only the material to be designated as core, but the degree of core to be applied. It can be predicted that teams will find the core concept initially frustrating and lacking in established, easy-to-follow signposts. It is here that the judgement and experience of the cataloger will play a large and crucial role. Most teams are accustomed to receiving the bulk of their material with decisions already made or to having the option of asking for a reconsideration, although some staff recently have been given training in this area by the selection officer. Teams/divisions should draw up individual team guidelines, proposals, and workflow for core in conjunction with reference/selection staff. Initially these guidelines need not be detailed or specific, but should spell out broad concepts. Only with experience will teams be able to draw up formal guidelines to be followed in the application of core.
Teams were invited to respond, either as a team or as individuals or both, to a series of questions included in the materials issued as part of the experiment. The following is a synthesis of the team responses prepared by Paul Maher and some specific comments selected by him.
All the teams regard core as viable for certain types of material, but not for everything. Most teams agreed that core is more efficient.
Hebraica: "Some savings could be realized with minimal loss of access" for targeted materials. GSL: Core provides "simplification of cataloging procedures coupled with adequate access." Music: With added flexibility in cataloger judgment it is a positive option.
If not, what would the team propose to improve it?
Not all teams suggested improvements. Those who did focused on training and documentation.
GSL: The only improvement we would suggest would be some sort of educational effort for catalogers in general that would make the principles and procedures of the core cataloging approach stand out in much the same way as the principles and procedures of MLC and copy cataloging stand out now. SSCD: Documentation needs to be improved to improve efficiency. ... [the] concept of core cataloging is fine IF it is based on catalogers exercising informed judgment ...
Core Cataloging Approach: Was the team able to incorporate into its cataloging outlook the core cataloging approach which emphasizes adequate, or even in some cases, minimal, access, in contrast to full level cataloging that calls for being comprehensive in every respect?
None of the teams reported major difficulties in incorporating core into its cataloging. However, "the meaning of core" seems to vary from team to team.
Hebraica: As long as cataloger judgment could be applied to selectively upgrade elements of the record, team members were satisfied with the level of treatment. ASCD: There was very little difference between subject cataloging for full and core ... the big difference was in the descriptive elements. GSL: we did not necessarily make any judgments as [to] the value of the selected items. We found that core treatment gave perfectly adequate access. SSCD: They found little, if any difference, in how they treated the books. If modified to reflect their ideas of essential, pared-down access, there would be a difference, however.
Data elements: Are there any data elements that the team would add to or delete from the complement used in the experiment?
None of the teams suggested the elimination of data elements per se. One team noted that remembering *not* to do something takes as much effort as just simply doing it, especially when macros are available.
Guidelines for core candidates: Was the team able to reach consensus on guidelines for what material would be a candidate for core level cataloging?
Some teams were able to reach a consensus and others not. Most agreed that the material included in the experiment did not reflect 'normal' work. The teams with larger arrearages reported ease of consensus on what to include.
GSL: Easy because of relatively large backlog of items. ASCD: Yes, because of large Slavic arrearage which otherwise might not still be cataloged. Music: Team unable to reach consensus for what is not appropriate for core, although they did identify brief score and brief sound recordings as being inappropriate. Hebraica: Did not represent usual mix of materials in Hebraica. HLCD: Team's materials ... [did] not represent normal workflow for team.
Special groups of materials: Did the team identify groups of materials that lend themselves particularly to core level treatment?
Again, there was a mixed response. In some cases, the perspective varied by cataloger's judgment (Hebraica), while one team felt an entire arrearage could be done as core (ASCD), and another team "did not apply any judgments as to value or suitability of materials for core" (GSL).
Basic MLC: In the team's view are there some materials within its scope that deserve no more than basic MLC in all cases?
None of the teams identified materials which deserve no more than basic MLC in all cases.
Priorities: Can the team extrapolate from its experience in the experiment the application of the concept of core level across all priorities in regular cataloging production mode?
GSL and SSCD view core as a useful option for all priorities. Hebraica would prefer to exclude priority 2 materials from core, while ASCD was hesitant to make a commitment.
SSCD: Yes, question really is 'What materials should NOT be done core, not the other way around.' ASCD: Not sure, certainly for Slavic materials, and would prefer core to MLC for other subjects in team.
Value of Materials: Did the team become comfortable with the new dimension of applying different levels of cataloging depending upon a judgment of the value of materials?
Most teams had no problem with relating the level of cataloging to the value of the material cataloged, at least for certain specific categories of material. Hebraica felt it would need to "continue to work closely with the selection officer to develop a better understanding of reader needs. ... Requiring catalogers to make on-the-spot decisions of whether to apply core or full regardless of priority will contribute further to the confusion."
Advantages/disadvantages: Were the concept of core level cataloging to be implemented at LC across the board, are there any particular advantages or disadvantages from the teams' view based on its experience in the experiment?
There was a very mixed response to this question. GSL saw no particular advantages or disadvantages, while ASCD saw advantages in the descriptive work, but not in the subject. SSCD viewed core as an opportunity to exercise flexibility and professional judgment while at the same time noting that "some staff will not accept loss of access--period." Hebraica saw core as a preferable alternative to MLC, but noted that a primary "drawback to core is difficulty of incorporating yet another level of cataloging, with fluid standards. Absent team consensus and selection staff guidance, leaving decision up to cataloger could be problematic." HLCD expressed the idea that core "serves no purpose for catalogers; does not meet reference needs."
Recording hours: does the team have any observations on the recording of hours?
Neither Hebraica nor GSL had any special observations regarding the recording of hours, except for noting additional record keeping. All the teams thought that accurate time-keeping was problematic at best. Both Music and SSCD suggested simplifying the categories of hours for types of cataloging.