Mining and Analyzing Circulation and ILL Data for Informed Collection Development

Poster: Curtin, Shane

Link, F. l., Tosaka, Y. t., & Weng, C. W. (2015). Mining and Analyzing Circulation and ILL Data for Informed Collection Development. College & Research Libraries, 76(6), 740-755.


Summary:

This article describes how one library tried reviewed ILL usage and  borrowing statistics for their academic journals in an attempt to asses the strength of their own collection. They drew on circulation data for each LOC class and compared it across a variety of factors. The results were more complex than they had expected. The question posed by their results was not what specific items to buy, but what subject areas to buy for.  They also realized that user preferences for different material types (such as digital journals versus print journals) could not be inferred through the aforementioned methods of study, since people will tend to order whatever is available. The researchers concluded by proposing a further study to address this question, and to establish benchmarks for collection use as it relates to purchasing. How much circulation should be expected for every dollar spent?


Analysis:

In my own library the ILL librarian does not fraternize much with the selectors. The number of ILLs we receive is small and seldom do we get more than one request per item in any considerable span of time. Due to our library’s membership in LINK+ we already have access to the catalogs of most Californa and Nevada libraries; most items can be obtained this way. We DO look at Link+ circ stats to see if we should order something, but not at ILL stats. Most ILL orders are for rare and old-out of print items anyway, not thing we could get our hands on even if we wanted to. Even so, this study was interesting vis-a-vis the content of our class. At the end of the study the author noted that the study is not to be taken as just an investigation of their own circulation, but as a parable of the effectiveness of using circulation stats as  collection development metric.

They of course, used LC call numbers as subject identifiers. While this is not the method recommended by this class, it is the only one available to most libraries, lest they choose to tag their records by subject collection, as we have been doing. In order for circulation data to be valuable to collection development in this sort of arrangement, a new field of some sort would have to be added in item records, denoting their collection grouping. An alternate method that occurs to me is to try and mine the subject data from the MARC 650 fields, but the variety of subject terms are much to numerous to  make this an effective strategy. If we escape the rigid methods of classic collection development by Dewey and item type, other methods of identification will have to be devised to make circulation data sortable in way that is useful  to selectors.

I was amused by one collection use measurement tool the author mentioned- the CFQ (Collection Failure Quotient) - a ratio of a library’s ILL borrowings to holdings. I think my own library is doing ok.