Effective Selectors? Interlibrary Loan Patrons as Monograph Purchasers: A Comparative Examination of Price and Circulation-Related Performance

Abstract

In the library literature, a great deal of interest in patron-driven collection development has recently been expressed, especially in those programs that link acquisitions with interlibrary loan. However, the implementation of such programs has been limited, at least in part because of concerns over the potential for wasteful spending. The authors will attempt to address this common concern by assessing whether monies spent via a patron-driven acquisitions program were more or less effective than monies spent via traditional modes of acquisition.

KEYWORDS:

• patron-driven acquisitions
• purchase-on-demand
• interlibrary loan
• use study
• monographs

Notes

1. As is frequently the case with a study of this sort, the law library was excluded from consideration.

2. The number of LC subclasses reported in an earlier study employing this dataset was 257. This discrepancy, and any other discrepancies between the two studies, was the result of a single ILL POD book that was misclassed as CM rather than CN. In the interval between the two studies, the cataloging staff was able to track down the rogue book and re-class it.

3. Prior to beginning the analysis of relative performance, the authors used SAS to test the distribution of book circulation rates (i.e., their annual turnover rates) for normalcy and then to perform a two-way factorial analysis of the dataset to determine whether, with respect to circulation as expressed via annual turnover rates, there were in the original dataset interaction effects attributable to means of acquisition and to LC subclass. In order to reduce noise effects that could have been caused by LC subclasses that did not have ILL acquisitions, the authors elected to use the subset of LC subclasses with circulated ILL acquisitions (137 subclasses; 66,396 books). The authors would tentatively offer the model constructed to fit the data as follows: Annual Turnover Rate = Acquisition Type + LC Subclass + (Acquisition Type x LC Subclasses). The model from the two-way factorial experiment was as shown below,

Where: μ is the overall mean response, τ_i is the effect due to the i-th level of factor Acquisition-Type, β_j is the effect due to the j-th level of factor LC Subclass, γ_ij is the effect due to any interaction between the i-th level of Acquisition Type and the j-th level of LC Subclass, and ϵ_ijk is the error which is identically and independently distributed with a zero median (note: LC Subclass had 137 levels; Acquisition Type had five levels). We tentatively concluded that, with respect to rates of annual turnover, the effect of LC Subclass depends on the level of Acquisition Type and that the interaction effect was significant since the p-value for testing the interaction effect was < .0001. Thus, it seemed prudent to include only those LC subclasses with ILL POD orders in the performance comparisons to follow. However, the authors would offer this model, again, very tentatively, as we feel that we are insufficiently familiar with the technique employed (i.e., a two-way factorial analysis of distribution-free, transformed ranked data from a non-normal dataset which used an alternative to the additive linear model) to know whether it may be novel or sufficiently validated (Sprent & Smeeton, 2007). In fact, in scanning the literature for examples, we were uncertain as to whether this method even has its own proper name, although something similar does appear to have been advocated by Hettmansperger & Elmore (2002).

4. What any particular library may deem an “acceptable” value for these ratios is subjective, but for this study, the authors took into account the books’ average annual turnover rate (1 circulation every 2.38 years) and the fact that a supermajority of the books (70.4%) had been available for checkout for at least that long. We concluded that the books in the topical areas should have been checked out at least once, on average. Thus, to be “acceptable,” their average book price and their average price paid per circulation should be fairly nearly equal.

5. One might be inclined, at this point, to glance at the values graphed in Figures 2, 3, and 4 and to conclude that Arts & Humanities, Social Sciences, and Sciences & Technology all performed roughly similarly. However, the reader ought to keep in mind that the ratios reported were for performance within the topic groups. The ratios may mask differences in performance that result from differences in average price paid within the topic groups, hence the reporting of ratios relative to average price paid in the text. The Sciences & Technology group's absolute performance (i.e., its price paid per circulation) may appear to be roughly comparable to the Arts & Humanities group's in several instances, but the performance relative to average prices paid for the books in some Sciences & Technology topic groups was actually better because the average prices for Sciences & Technology books were higher. This struck the authors as noteworthy and as a bit of a surprise because the arts and humanities are generally considered more book-oriented and the STEM fields more journal-oriented.