Monday, April 11, 2011

The Cost of Deselection (6): In-Stack Review

For some reason, deselection decisions seem more difficult than selection decisions. The thought processes are similar, but the consequences are different. But are they really? In selection, the worst-case scenario is not selection of a book that is never used, but rather non-selection of a book that is later wanted but no longer available. The best-case scenario is selecting a book that is ultimately used many times before anyone knows that it is wanted. In deselection, the worst-case scenario is the removal of a book that is subsequently wanted but no longer available--very similar to the worst-case scenario for selection. The best-case scenario is removal of books that will never be used, and retention of those that are subsequently used.

But there remains a perceived finality to the deselection decision. That perception is in many respects false, as it is perhaps easier to re-obtain content now than at any time in history. Irreversible mistakes are rare. But the stakes seem high, and for now it is rare indeed to meet a selector who will withdraw a book without looking at it. In the previous post, we considered deselection from lists. Here we look at the costs associated with book-in-hand review. There are two approaches, each with slightly different time investments: in-stack review and staged review. Today we consider the first of these:

In-stack review:

Some rights reserved by Andrew|W
In this model, books are reviewed in situ. While it is possible for a selector to use a list arranged in call number order for this process, it is more common that staff or student workers find and mark withdrawal candidates in the stacks, either by applying colored tape or turning the books down. This carries the risk of users disrupting some indicators, but also confines searching and error correction to hourly workers.

1. Marking the Candidates: Again, let's assume that a student worker, using a list arranged in shelflist order, can find an label 1 title per minute. This may seem slow, but this average has to incorporate looking for items that are missing or misplaced, annotating the list, and moving or marking found items. In eight hours, 480 books could be marked. Let's round up to 500/day.

To mark our entire list of 10,000 withdrawal candidates, then, would require 20 person-days at 500/day. If we assume an 8-hour day and apply a student wage of $10/hour, the process of finding and marking the withdrawal candidates is about $1,600.

2. Librarian/Faculty review: Once the candidates are marked, librarians or faculty members can enter the stacks, either with or without a list, to review and decide on withdrawal. While finding them will be easy, the decision-making will almost certainly be slower than the marking process. Deselectors will look at context, at nearby items that cover related topics. They may choose to consult the catalog to see what other books are held by the author. They will need to indicate their decision in some way, and perhaps provide a reason for retention.

Let's assume these thought processes and actions take twice as long as the marking; i.e., allocate an average of about 2 minutes per title. That equates to 250/day. Our 10,000 titles would require a total of 40 person-days. If we assume an 8-hour day at $35/hour, the cost of in-stack review by deselectors is approximately $11,200. ($280/person-day x 40 days).  

The total cost for in-stack review:
  • $1,600 for marking
  • $11,200 for deselection decisions
  • $12,800 Total, or
  • $1.28 per title
 Links to related posts:


    1. Hi Rick,
      Just found your excellent blog. One question that I'm asked often is "Do you weed your ebook collection?" Generally my response is "No, why would we do that?" So Rick, can you tell me? Why would we do that? Are there reasons (thinking through cost/benefit) that we should weed ebooks? [We do in fact weed ebooks very selectively as books that are totally out of scope come to our awareness. ie knitting for dummies, etc. particularly in our PDA collection.)

    2. Hi Deb,

      Interesting question, and one I haven't thought much about yet, since those ebooks don't take up much space! A couple of comments, though:

      1. I've started to think about collection development (especially in the age of PDA) as "curating a discovery environment" rather than building and managing a collection. Sounds as if you're doing this when you're selectively loading records for your PDA titles.

      2. Curating a discovery environment, in my mind, would imply periodic review of usage statistics and perhaps corresponding removal of records for no/low-use titles.

      3. It's hard to see what the payoff is, though. We already know we do a poor job of predicting what users want. What's the harm of leaving a no-use record in the catalog and available for subsequent discovery? I guess a huge number of these would eventually affect indexing speed, but...

      4. It would be hard to see ebook weeding as a priority, unless the bib data is bad or unless there are charges associated with keeping the record "live."

      That's my 2 cents' worth, anyway!