Focus on Assessment - Aug 29, 2011

It is not unusual for a library borrower to reply to an overdue notice with the statement that the item in question has already been returned. Because we take pride in our work it is natural to feel defensive in light of such a claim, but we must realize that occasionally something slips by us. Humans make errors and checking in returns is a very human-centered process. To mitigate such errors some libraries actually engage in the time-consuming practice of double discharging returned books rather than risk making such an error. You have to wonder if their error rate is so high that it is worth executing such a high-volume task twice? And you might wonder, “What is a reasonable error rate for manual processing?” Is it 1 or 2 percent — 2 errors out of every 100 items handled? For a library, like Penn State that collectively discharges 850,000 books annually that number would be around 17,000 and that seems like a lot of errors!

By acknowledging our own fallibility, taking each claim seriously, and following a standard process to investigate and respond to the borrower, we raise our users’ confidence in our professionalism and integrity. By establishing an error rate for your unit you may raise staff confidence in their accuracy and you can respond with conviction when you let the reader know that the book was not returned. Of course, if a book is reshelved without being discharged then an apology is in order. “We’re sorry we do miss one once in a while. Thank you for allowing us to correct the error.” Fines would be adjusted accordingly.

How do you establish an error rate? Simply divide the number of actual errors by the number of errors added to the number of discharges: Error Rate = Errors/(Errors+ Discharges)

Not every claim means that the book was in fact returned so it’s really not fair to say that a claim equals an error. The first step in verifying a returned claim is to check the shelf to see if the book is there. For our purpose, an error is represented by a book sitting on the shelf that was not checked in.

For the calendar year 2010 Lending Services at Pattee/Paterno logged 308 returned claims for 387,000 discharges, 137 of those were substantiated errors or 4/10,000. While not an ideal number, it does tell us that our accuracy rate is a pretty respectable 9,996 out of 10,000. Because we monitor this on a monthly basis we know that during peak periods such as May and December errors rise with the number of check-ins and claims rise with them. Understanding that allows us to consider how to address the problem, perhaps with renewed attention to the work, fewer distractions, a more measured pace or more hands on the work. The reason for assessing errors in the first place is to improve the process.

In reviewing the numbers we learn that 45 percent of the 2010 claims (136) were resolved by the item being found on the shelf. Another 20 percent were returned to the library, after the claim was reported by the borrower, usually within a day or two. The rest are eventually resolved when the user locates and returns the book or pays the library for the lost item.

Relevant reading: Reducing Check-In Errors
at Brigham Young University Through Statistical
Process Control, Spackman, N Andrew,
Journal of Access Services. Vol. 3, no. 3, pp.
1-16. 2005