Saturday, January 14, 2017

How are my numbers?

One stereotype about librarians that bothers me is that they can't do math. I was a statistics minor in college and that has been really valuable in my day-to-day work as a librarian. Here's an example of how knowing about statistics helps me:

I had the bright idea today that I would compare how my later purchases for the collection did compared to my early ones in order to see if I had learned anything from my earlier work. I thought I would compare the average number of checkouts per month for the batch of titles I ordered from March through June with the batch of titles I ordered from July through December.

Here's what I found for the average checkout per month for each area of nonfiction, by Dewey number:
000s (computers): 0.44 checkouts/month
100s (self-help, philosophy, psychology): 0.69 checkouts/month
700s (crafts, art, music, film, sports, home): 0.57 checkouts/month

I wanted to go back and check these numbers against the same metric for my earlier picks, but when I tried that I realized I had a big problem: We keep our items on a special "New Items" shelf right near the front of the library for the first three months after we've acquired them. This spot is way more visible than the rest of nonfiction so it's common for items to circulate significantly more heavily in the first three months compared to any random three months later in their 'lifetime' in the library collection. With the two batches I wanted to compare, the books I made a list of in June have been in New shelving for 3 months and in regular shelving from anywhere from about 3-6 months. The later batch that I made a list of today has been in New shelving for up to 3 months and in regular shelving from anywhere from no time at all to about 3 months. So of course the averages for the latter will be higher--they are still in their high-checkout honeymoon period where they are being shown off on the New shelves. When I checked my older batch of 000s and got an average of 0.14 checkouts a month, got confirmation of the problems that the existence of the 'New' shelving was creating for the analysis (and got discouraged by those very low numbers!).

However, finding those numbers wasn't a waste of time, even though I was unable to compare them to my earlier selections as I had hoped. I was happy to see that the three numbers were fairly close to each other--in all three of my subject areas, the 'typical' item goes out about once every two months. I'd say that a typical checkout period at our library is a little over two weeks (the loan period is longer, but what I'm thinking about is how long a book is actually absent from the library when it is checked out), so that means the titles I have ordered are spending about 3/4 of their time on our shelves and 1/4 of their time out with patrons. Intuitively, I feel like something between 1/2 and 2/3 of the time is how much I would like them to spend on the shelf (I didn't realize I had an intuition about that, I guess I learned something about myself today), and this is not too much over that. I also use "less than 1 or 2 checkouts per year" as a baseline for whether to consider discarding something, and the titles that I ordered are comfortably above that margin so far.


Overall, I give myself 4 out of 5 for this attempt.

5 comments:

  1. Libraries would be better if their policies were based on good data rather than on what some middle-class white woman has been doing for 30 years

    ReplyDelete
  2. It's a good thing you caught the "New Books" problem...it's easy to miss a thing like that, you get so caught up in doing the numbers.

    ReplyDelete
  3. have you found good stats to compare yours with?

    ReplyDelete
    Replies
    1. Not yet. I think I'll just have to wait a few months and re-run the same tests when the 'New' boost has been more diluted. I have a general sense that I'm getting better, though.

      Delete