Measuring Collection Use

Despite Neil Young's warning that 'numbers add up to nothing', sometimes they have to suffice. In the data gathering that underpins our deselection and shared print projects, we at SCS spend a lot of time looking closely at circulation statistics. Several factors influence what can be gleaned here:

•  ILS: Different library systems capture and store circulation data in different ways.

• Duration: Most libraries retain circulation data back as far as their last system migration, though a small percentage port historical checkouts over as part of the data transfer process.

• Definitions: Often a checkout is just a checkout. This usually includes direct borrowing and ILL transactions, but not always. Some libraries also use checkouts to monitor workflows, charging books out to Acquisitions, Cataloging, Bindery, etc.

• Transaction dates: Some systems capture only circulation totals. In others, it is possible to learn the date of the most recent circulation transaction.

Not to put too fine a point on it, there are no standards for circulation data. While this imposes some limits on the analysis that can be done for an individual library, it can be even more problematic in shared print projects, where it is necessary to derive a common basis for circulation activity. Even with its limitations, though, circulation data provides an important objective measure of collection use.

No one would argue that circulation activity constitutes the full picture, of course. As Walt Crawford and Michael Gorman point out in Future Libraries: Dreams, Madness and Reality: "To be effective politically, it is vital to record the totality of collection use." [WorldCat record]. This is easily confirmed in conversation with faculty and other library users, who often assert with some vigor that circulation represents only a partial picture of use. They base this, reasonably enough, on their own habits, and there is some quantitative support for this in cases where statistics for in-house use are available.

In one library we worked with recently, a 2-month sample of reshelving counts indicated that ten in-house uses occurred for every circulation. That is the highest number we have encountered. A more commonly-reported level is echoed in Future Libraries: "When libraries have counted in-library use, usually omitting pure browsing, the numbers are 2-3 times as high as actual circulations."

Especially when deselection is being considered, recognition of any and all use is important. The last thing we want to do is remove something from the shelves that is actually used. How can we be certain we're getting the fullest picture? 

The obvious answer is to institute or re-institute reshelving counts. This is a frustrating conclusion. It seems completely counter-intuitive and backward-looking to add this sort of work to a library's daily operations in the digital era. Reshelving counts are time-consuming, and require cooperation from users. They shift the focus to print when all signs point toward declining use and value.

But we need the data. We are beginning to make long-term decisions about the future of print collections, and we need those decisions to be as informed as possible. This means we need to capture in-house use--at least in some form.

The simplest and most accurate approach is to provide and promote regular reshelving counts--to identify and count all books used in the library. Stacks workers can scan the barcode and either tally in-house use separately or count it as a checkout. This approach ties use statistics to specific titles, which is optimal.

Some libraries will find this investment hard to justify or sustain. In those cases, a sampling approach could be adopted. Adopt reshelving counts for one week per quarter or one week per semester. This less labor-intensive approach could provide sufficient data for extrapolation; i.e., to calculate an estimated rate of in-house use that could augment actual circulation statistics.