Archives For data analysis

book and skull with title The Almost Dead Experiment

What’s a rational weeding philosophy? When is a book really dead? 6 months? 2 years? 4 years? More importantly: how do we know? Progressive, cool libraries stick to a 180 day weeding target, while old, stodgy libraries pile books to the rafters (’cause hey . . . we got the space). And some libraries (ahem, East Lake County) hate weeding so much and so blindly that they create a fake patron account to “save” books. All of these beg the same question: what data are these decisions based on?

In my new position managing the adult collection at KPL, I’m in a complex situation. While we tend to have a more conservative weeding target (4 years at main downtown library), I personally found myself leaning towards the more aggressive end of the spectrum – I just assumed that if a book sits on the shelf for 2 or 3 years, it’s probably not getting checked out.

Until now.

CollectionHQ, an evidence-based collection management tool designed for libraries (recently acquired by Baker and Taylor), is normally used for weeding and selecting. But my favorite tool, called “experimental placement,” allows you to track particular books or collections and see how they perform over time. I ran an experiment 7 months ago, the results of which completely blow my mind. It turns out that, at my library, books that haven’t been checked out in three years – three years! – are still not dead yet.

Looking at the Dewey ranges 000-550, I was able to find 907 books that hadn’t checked out in three years (our weeding target is currently set at four years). I put those 907 Almost Dead Books into a CollectionHQ experiment, and I waited.

Out of 907 books in the experiment, a whopping 232 (25%) were subsequently checked out by patrons. That is to say, 1 in every 4 Almost Dead Books were checked out. And of those 232 that were checked out, many were renewed multiple times and/or checked out again by someone else – making a net total of 469 circulations from Almost Dead Books.

Before the experiment, my prediction was that 1-5% of the Almost Dead Books would circulate again. I had to double check the numbers. I even checked the ILS at most of these titles, just to be sure actual real physical people were checking them out. Indeed.

I predict the second half of the Dewey ranges would do markedly better (550-999: diet, workout, cooking, business, crafts, travel).

This data sort of justifies our “conservative” target – clearly people still want some of these books. You could argue that weeding Almost Dead Books would actually lead to more total circulation by freeing up space for newer books.

I don’t want to oversimplify things. Weeding isn’t a simple 6 months vs. 4 years decision. Every Dewey range and every genre of fiction should  be treated differently, and there are myriad other considerations. I just wanted to repeat the buzzword because it matters so much in our profession: data-driven decision making. Whatever you do at your library, try to get data to inform the decision. You might be surprised.

Questions about how to do these experiments in CollectionHQ?

A colleague sent me the recent Ithaka S+R report on Library Leadership in the Digital Age by Deanna Marcum, commenting that it was a wake up call for what we’ll have to do as leaders in the near future.

I was delighted to find that the report is filled with concrete, actionable suggestions. Sometimes reports with “21st Century” or “Digital Leadership” in the title can be aspirational yet vague. In this case, Marcum’s recommendations are all grounded in reality. She calls for skill building across the library (both in leaders, and the skills they need to foster in their staff) and for a focus on user needs.

For instance, the report calls for leaders to pay attention to business fundamentals. This recommendation hit home for me. I’ve long felt at sea when it comes to large budget management. After hearing the results of Douglas Crane’s interviews with library directors, I realized that I need to focus my professional development this year on business principles – so much so that I’m considering MBA classes.

Marcum says that digital leaders need the clarity of vision to help employees with traditional skill sets navigate a landscape that still values those skills, and also demands new abilities:

Digital leaders will help the more experienced staff move into the digital world with confidence and enthusiasm.

Confidence and enthusiasm! What a goal. I’m now considering how I might build my team’s skill set so that we can react to changes with positivity rather than fear – and seeing that we need data analysis skills, stat (even if we start with pivot tables).

Check out the full report on the Ithaka S+R site. Do the recommendations resonate with you? What skills do you think digital leaders need?