The RLG Conoco Study and its Aftermath: -Is Resource Sharing in Limbo

Faculty members at academic libraries have come to expect certain titles to be on the shelves no matter how expensive they are. Their philosophy seems to be: "Any decent library should own that title." In some cases, the title in question has little prospect of being used- possessing it is a point of academic honor. This type of possession, however, is an honor we can no longer afford. Library budgets demand that we forego purchasing seldom-used books when borrowing them is more cost-effective.

Borrowing in turn calls for the certainty that a desired item can be readily located and that permission to borrow will be given. Resource sharing and cooperative collection development are, as a result, interconnected goals of library consortia that want to pursue common goals. When one member of a consortium can rely on another member for a given set of library materials, that member can limit purchases on the local level while still providing materials to users. Thus, cooperative collection development, with its coordinated scheme of purchasing and non-purchasing, lies at the heart of resource sharing.

Without planning and coordination, collection gaps and overlaps would prove insurmountable. A collection gap at one library can be accommodated if the same collection gap does not exist at a cooperating library. Collection gaps are hindrances to cooperation only to the extent they exist system-wide. Collection overlaps, those titles that more than one library has acquired, may or may not represent a waste of funds. Overlaps may signal necessary duplication if the titles are heavily used; if overlaps are rarely used items, however, cooperative collection development can save money with minimal inconvenience to users.

Still, the idealistic concept of sharing academic resources, for all its promised benefits, seems likely to share the fate of Eastern European socialism. The theory of communal good may get only a figurative pat on the head unless an institution can see immediate self-interests being served. The luxuries of acquisitiveness and local access, even in times of dire financial shortages, appear deep-rooted and habitual. And questionable systems of document delivery only make selectors more reluctant to ask their faculty to wait for materials available elsewhere.

The pre- and post-histories of the RLG Conoco Study serve here to put the impulses toward cooperative collection development in context. Within this consortium, as within the academic library world as a whole, the virtues of such cooperation have been more often proclaimed than implemented. Patterns of resource sharing-if there are any-may become clearer as we see what has been preached versus what has actually been practiced.

The cooperative movement in libraries extends back to the 19th century. The original efforts concentrated on bibliographic descriptions as found in union lists, with no mention of cooperative acquisitions.(1) The Farmington Plan of 1947, the first major experiment in cooperative collection development, was considered a failure in the long run. (2)

One librarian in 1957 felt that academic libraries of that time were obliged to cooperate at regional and state levels, but that regional agreements would be difficult to draft until libraries better knew their own areas of strength and weakness.(3) The following year, another author noted that smaller institutions tended to rely more on larger institutions than on each other for interlibrary loan, which placed a disproportionate burden on the former. Hence, a "will to cooperate" would have to come from all sides for mutual benefit.(4) In 1959 the director of the Stanford libraries carried forward the same theme, stating that the resource functions of the larger libraries (with the possible exception of the Library of Congress) might profitably be transferred to regional centers such as the Midwest Inter-Library Center- a center that later developed into the Center for Research Libraries.(5)

Despite such pronouncements of obligation and altruism, it was still clear that cooperation would continue to take a back seat to institutional ego. A study in 1963 found a strong positive correlation between the size of a library and the academic ranking of the institution it served.(6) From that standpoint, the notion of cooperative acquisitions seemed an invitation to academic mediocrity. The post-Sputnik funding boom of the 1960s, with its resultant collecting fever, left individual college and university libraries with neither the will nor the need for cooperative acquisitions.

In the 1970s the situation began changing again toward a reexamination of cooperation, driven by the information explosion, by inflation, by the demise of budgetary largess, and by burgeoning automated networks and consortia.(7) In this new atmosphere, the earlier lessons of unequal "partnerships" were not forgotten, and the point was made that "cooperation cannot be based on institutional altruism."(8) Conversely, a cooperative system would never thrive with libraries trying to get something for nothing. (9) Resource sharing could never operate well until local goals and consortium goals coincided for all members of the partnership. There were legal and fiscal barriers to cooperation (10) as well as psychological ones.(11)

The tight budgets of the 1980s forced even greater selectivity on libraries, and resource sharing reached new heights-- at times transcending the level of planning and discussion. In a relative sense, the past decade could be called the "Golden Age of Resource Sharing" or at least the most golden so far. The pool of shareable materials, however, did not materialize automatically. In order to be successful, selectors had to follow a careful collection development model (12) and calculated changing trends in research needs, in financial fluctuations, in the publishing industry, and in automation.(13)

When libraries acquired the same materials ("mirror-collected") to a high degree, then cooperative collecting became impossible.(14) It happened that "all too frequently, cooperation [was] merely a pooling of poverty,"(15) especially when emphasis was put on theoretically "balanced" collections that resembled most other "balanced" collections.(16) Pooling resources only worked when the pool was replenished,(17) otherwise, the questionable phenomenon of "overlap" was overshadowed by the irremediably darker curse of collection gaps or "underlap."(18)

Another barrier to cooperation was the time an item took to get from the lender to the borrower. A study of faculty and student users in 1979 indicated that 64 percent felt they needed their materials in two days or less, 23 percent would accept a response time of three or four days, while only 12 percent found a wait of one to two weeks to be acceptable.(19)

Selectors best filled specialized needs of library users by widening the focus of planning beyond the local level. To provide for shared resources, librarians began planning on a larger scale, such as through a state or other bibliographic consortium.(20)

With the Farmington Plan, Public Law 480, the Higher Education Act of 1965,(21) and other models of cooperative initiatives before it- particularly for acquiring foreign-language materials- the Research Libraries Group (RLG) embarked on similar ventures. One of the first discoveries of the RLG's founding libraries in 1978 was that there were no legal or financial precedents for the cooperative work they wished to do.(22)

Clearly, an important step toward practical resource sharing was the creation of a conspectus of each library's holdings, basically a map of strengths and weaknesses that could be viewed by all members of the group.(23) The Association of Research Libraries (ARL) expanded on this concept with the North American Collections Inventory Project.(24) Attempts to fine-tune the subjectivity of the resulting Conspectus have been made through verification and overlap studies. (25)

Another practical step was the allocation of "primary collecting responsibilities" for various subjects divided among member libraries.(26) RLG libraries have teamed with the Center for Research Libraries (CRL) and the Library of Congress' Public Law 480 program to reach a practical level of cooperative acquisitions for South Asia that may serve as a model for collecting in other foreign areas. (27)

The efforts of these consortia marked a step beyond cooperation to the more promising realms of collaboration.(28) The RLG Conoco Study, carried out in 1985, was one such effort.

The Conoco Study was designed to measure the possibilities for resource sharing among RLG members. RLG's Collection Management and Development Committee oversaw the study. The methodology of the Conoco Study involved three distinct steps:

Questions 2, 3, and 7 could be called "environmental factors." These were especially crucial, since they represent what might be changed through cooperation: the methods of acquiring titles, the relative difficulty of acquiring titles, and the effect of document delivery on selection decisions.(31)

The findings can be grouped according to 12 hypotheses formulated by the project team.(32) The hypotheses themselves may be categorized into three focuses: selector behavior for hypotheses 1 through 6, collection overlap for 7 through 9, and cost benefits for 10 through 12. These hypotheses follow, along with the findings that help to confirm or to refute them.

Hypothesis I (selector behavior).
Selectors will be especially willing to depend on other collections to meet the need for material thought to have potential research value.

The distinction made here is between materials purchased for demonstrated research needs and those with potential (but undemonstrated) research value. The items purchased for future research were hypothesized to be the most shareable (or borrowable). This hypothesis was confirmed. Both German and geology selectors apparently want to concentrate cooperative action on such research materials.(33)

Hypothesis 2 (selector behavior).
Selectors will want to focus cooperative action primarily on materials designated as belonging to research level collections.

This hypothesis, related to the first, was also confirmed. But while both German and geology selectors were in basic agreement about research materials--- considering more than 50 percent of their titles at the research level to be targets for cooperation-the German selectors considered only 17 percent of the titles at the instructional level to be shareable, while geology selectors pegged 52 percent of the titles at the same level as shareable. (These differences could not be explained from study data.) (34)

Hypothesis 3 (selector behavior).
Selectors will not wish to depend on other', collections to satisfy traditional library collection interests.

The survey refuted this hypothesis. German selectors categorized 19 percent of their titles as acquired for "traditional library interests unrelated to current programs/research interests," compared to only 10 percent of the geology selectors. Still, in both cases, the incidence of responses indicating that selectors would continue to purchase such items (thereby "not wishing to depend on other collections") was very close to the average for all categories of titles. Thus, those items satisfying "traditional library collection interests" were found to be as potentially shareable as any others.(35)

Hypothesis 4 (selector behavior).
The willingness of selectors to depend on cooperatively held collections will increase with a decrease in the number of working days needed for delivery.

This was also refuted by the data. In response to question 7, German selectors most frequently chose seven days for response time. Geology selectors, on the other hand, most frequently chose three days. Even so, there appeared to be another factor for both disciplines: "only the combined factors of a title's shareability coupled with strongly perceived potential research value (for German selectors) or qualities supporting faculty/research (for geology selectors) led selectors to opt for a shorter time." (36) The one-day delivery option (which required some suspension of disbelief on the part of the selector) did not show a linear correlation. That is, a quicker delivery period than three or seven days did not result in drastically different decisions about purchasing. (37)

Hypothesis 5 (selector behavior).
Selectors will be unwilling to depend on other collections to supply Material thought to be needed quickly by users.

Hypothesis 6 (selector behavior).
Selectors will be unwilling to depend on other collections to supply material heavily or frequently used.

The selectors adamantly supported both of these hypotheses. "The perception of heavy use or the compelling need of rapid access is virtually decisive among German selectors, and weighs heavily. for the geology selectors."(38) The German selectors indicated that they would repurchase 91 percent of the titles considered to be heavily used and 71 percent of those thought to be needed quickly. The geology selectors indicated that they would repurchase 73 percent of those titles deemed heavily used and 60 percent of those titles needed quickly. The differences found here might be traced to the fact that German and geology selectors apparently have a different notion of interpreting "quickly" (as indicated by the three-day versus seven-day delivery time required). The only category with higher repurchase percentages than "heavy use" and "needed quickly" was that of materials known to be needed or required for courses (98 percent among German selectors, 67 percent among geology selectors).

German selectors indicated that only 21 percent of the titles they examined could be classified as either heavily used or needed quickly. However, geology selectors gave the "heavy use" designation to 28 percent of the geology titles and the "needed quickly" designation to 8 percent, combining to 36 percent of the total. Both types of selectors agree on the necessity of having materials from these two categories on site and about the impractical nature of sharing them.

In attempting to explain why the perceptions of "heavy use" and "needed quickly" differed between selectors, the project team postulated that geology selectors tend to "perform a larger number of public service functions, and thus have a more intimate knowledge of their clientele than German selectors.(39) To test whether the differences are due more to perception than reality, it might be necessary to use comparative German and geology circulation statistics in the case of "heavy use," and perhaps a user study to determine relative frequencies of "needed quickly."(40)

Hypothesis 7 (collection overlap).
There is more homogeneity in science collections than in humanistic ones.

This hypothesis holds true in terms of the German and geology literature examined in this study, Overlap data showed that 73 percent of. the titles in the German literature sample were held by more than one institution, while the geology sample yielded 82 percent for the same measure. (41)

Four percent of the German titles were held by all libraries, and 19 percent were held uniquely. Put another way, German titles tended to be jointly held by a small (42) number of libraries. For geology, on the other hand, 14 percent of the titles were present in all of the collections, and when figures from the sole "instructional" level collection were eliminated from the data, the ratio jumped to 21 percent. Uniquely held geology titles amounted to only 9 percent. The converse of the German literature situation is true: geology titles are far more likely to be held jointly by a large number of institutions than by a small number. (43)

This greater homogeneity in geology collections might be attributed to the fact that they support a journal-based discipline while German literature relies more on monographs. A comparison with an earlier study in mathematical journals shows patterns very similar to those found here for geology. An earlier French literature study suggests a more gradual curve from "unique" to "held in common" than what was found in the German literature portion of the Conoco Study.(44)

A further consideration within German literature, however, is the fact that the unique titles tended to be belles lettres-the primary literature itself as opposed to the secondary or auxiliary literatures. When belles lettres were ignored, the German overlap of 83 percent was quite comparable to the 82 percent of the geology sample. Obviously though, primary texts are crucial in studying a literature. This is a complex area that German selectors might target for cooperation, especially when material deals with dialect literature, popular literature, and so on.(45)

Of uniquely held titles in German, the overwhelming majority were held by the New York Public Library. Unlike other libraries, it tended to acquire uniquely held titles through a detailed approval plan rather than by firm orders. (46)

There were positive signs of resource-sharing potential: "It has become apparent that there is a richness of RLG overlap in current materials that must be encouraging. It also highlights the broadness that needs to be captured in any cooperative effort."(47)

Hypothesis 8 (collection overlap).
Non-English-language Material is underrepresented in scientific collections.

This hypothesis held true in the study. In the geology sample, language turned out to be the single most statistically accurate predictor of whether or not a title was held. Foreign-language materials constituted only 22- percent of the entire sample, but 55 percent of all titles not held. Forty-five percent of the foreign-language titles were not held by any of the nine RLG libraries participating in the geology portion of the study. Russian-language materials came out the weakest, with fewer than half of the titles held. These are not surprising data, given the fact that several of the participants explicitly exclude foreign-language materials from the scope of their collection."(48)

In considering foreign-language publications in the German literature sample, investigators had proposed an unstated sub-hypothesis that materials from Switzerland, Austria, and East Germany, which are particularly difficult to obtain, would be under-represented. This appeared to be the case only for Swiss and Austrian materials. Only 60 percent of the Swiss and 66 percent of the Austrian materials were held by any of the participating libraries, while Fast German materials were held at the rate of 76 percent, one percentage point better than West Germany. Of the East German titles not held, 92 percent were belles lettres.(49)

The above fact led to an erroneous statement in the RLG summary of the Conoco Study:

Almost all of the titles not held in the German sample fell in the category
of belles lettres. 27 percent of the German sample was not
owned by any participating institution; 92 percent of these items were belles lettres.(50)

In fact, the 92 percent figure holds true only for East German materials; the figure for all German-language materials is 71 percent, a much more reasonable figure.(51)

Hypothesis 9 (collection overlap).
"Rely on others" will tend to be a default answer for those titles that are perceived to be within an institution's scope, but have not been chosen.

This hypothesis was refuted. Of the "do not hold" responses for German titles, 17 percent were considered "out of scope," 23 percent were "rely on others." In the German portion of the study, two other factors were also approximately equal: 21 percent of the titles had not been identified, but would otherwise have been acquired; and 22 percent were inexplicably not held.

For the geology sample, "out of scope" accounted for 30 percent of the unacquired titles and "'rely on others" for 32 percent. The "would buy if identified" category only accounted for 10 percent of the unacquired geology materials.(52)

The conclusion was drawn that "geology selectors would seem on the whole to be more confident that they are selecting materials in scope . . . and are more readily able to designate material 'out of scope' than German selectors."(53)

Hypothesis 10 (cost benefits).
High prices are rarely a reason for not Selecting material, and are therefore not in themselves a strong motive for cooperative action in collection management.

This hypothesis was borne out. The "too expensive" response for unacquired titles occurred rarely, only .2 percent of the time for German titles and .9 percent for the geology sample. The data showed that in all other cases expense was not an overriding, or even an influencing factor.(54) In fact, however, only one item in either sample exceeded $500 in price: a geology item with a price tag of $1,050.

Among the most revealing findings was a strong correlation (.29 for geology selectors and .46 for German selectors) between an item's cost and the likelihood of its being selected by a large number of institutions- i.e., "the more an item costs, the more likely it was to be held by a large number of institutions."(55) This correlation may not be as surprising as it first appears. More expensive items tend to be core-level materials produced by commercial publishing houses that charge more for their products than do government agencies, scholarly societies, and so forth. The cost of an item, within certain limits, may even serve as a psychological flag to the selector of the item's worth or centrality to the discipline. The larger advertising efforts made for such items-usually commercial publications-will also increase the probability that selectors become aware of their existence.

The less expensive items tended to be "endangered titles," i.e., those found in no libraries or in only one.(56) In geology, they tended to be foreign titles and those not in the mainstream trade market. For German literature, they were the ones coming from minor publishing houses and printed in small editions.

Hypothesis 11 (cost benefits).
Selectors spend very little to satisfy consortial or other contractual obligations.

This hypothesis was borne out, with the exception of one institution. The German selectors identified 21 titles as having been purchased to satisfy consortial obligations; the geology selectors identified only 3. Of the 21 German titles purchased for the consortium, 19 of them were at the expense of Stanford. The implications seem to be (1) that "the much heralded charge of cooperation has not yet spawned any more than minimal activity," and (2) that, with one exception, RLG libraries do not seem to have 11 any economic stake in consortial or other cooperative resource sharing."(57)

Hypothesis 12 (cost benefits).
Cooperative collection management will produce significant staff cost savings by the avoidance of duplicated selection and acquisition work for titles that are difficult to identify or maintain.

The study revealed a significant difference between selectors' opinions on the difficulty of acquiring and maintaining titles from the sample (as indicated by their positive responses to the second, fourth, and fifth choices in question 3). This difficulty was cited for only 6 percent of the German titles, while it was indicated for over 25 percent of the geology titles. The primary reliance of geology researchers on journals rather than monographic literature must be taken into account in this variance. And even in the case of geology, there was a range from 4 percent (at Colorado State, with an instructional collection) to 52 percent (at Stanford, with a heavy research emphasis).

Thus, the hypothesis proved to be true for the geology collections and not applicable to German literature. However, the project team calculated that if the personnel costs of identifying and maintaining each title considered "difficult" were partitioned in an organized manner among the consortium members, substantial overall savings would result.(58)