Academic
A 1973 study in the Art and Architecture library at Washington University involving a collection of 50,202 volumes and lacking any exit control (sample size: 594 volumes) showed that153 or 25.5% of the collection was missing. Additional searches during a subsequent 8-month period reduced the loss rate to 55 (9.2%), while by the end of a 2-year period the loss rate had dropped to 33 (5.5%) volumes. At the 95% confidence level, this adjusted loss rate ranged between 8% and 4% of the collection. (Source)
A study reported in 1976 at Washington University’s central library (Olin Library) involving book loss in a collection of 850,000 volumes protected by a manual check of exiting patrons (sample size: 2,949 volumes) showed thatadditional searches during a subsequent 10-month period located almost half of the missing volumes, leaving only 86 (2.91%) of the collection missing. At the 95% confidence level, this adjusted loss rate ranged between 2.7% and 1.7% of the collection. (Source)
Ibid…. showed that161 (5.45%) volumes could not be found. With a 95% confidence level, the loss rate ranged between 6.3% and 4.7% of the collection. (Source)
A study reported in 1977 of book loss at Northwestern University showed thatsampling to determine book loss may provide an artifically high loss figure. An inventory of 110,000 high-use books at Northwestern University revealed a loss figure of 1.8%; a sample of the same collection indicated a loss rate between 2.2 and 4.0% with a 95% confidence interval. (Source)
A study reported in 1982 at Washington State University of a relatively homogeneous book collection (a recent social science collection of 100,000 volumes) showed thatpatron reports of missing items in a collection subject area were the strongest single predictor of the actual number of missing items in that area with an r2 = .4049 (significant at the .01 level). (Source)
Ibid…. showed thatcirculation rate of a collection subject area was the second strongest predictor of the actual number of missing items in that area with an r2 = .3187 (significant at the .01 level). (Source)
Ibid…. showed thatcombining patron reports of missing items in a collection subject area with the circulation rate of that subject area produced a combined predictor of the actual number of missing items in that area that was a more powerful predictor than either of the 2 factors singly with an r2 = .4892 (significant at the .01 level). (Source)
Public
The experience of the Huston Public Library in inventorying its collection (236,519 titles) showed thata 2-man team comparing the shelflist card to item on the shelf could check 42 volumes per man per minute. (Source)
Ibid…. showed thatcatalog cards could be pulled from the catalog at the rate of 40 cards per hour per person. (Source)
A full inventory of the Main Library of the Huston Public Library after a sample had been used to re-estimate missing volumes in a collection of 236,519 titles showed thatthe sample overestimated volume losses by 10% (sample projected losses = 41% vols.; actual losses = 31% vols.) and title losses by 15% (sample projected losses = 35%; actual losses = 20%). The discrepancy was felt to be due to using many individuals unfamiliar with the shelflist to take the sample rather than using a single knowledgeable invididual. (Source)
