Note: Lydia Porter, an office assistant in the J. Willard Marriott Library at the University of Utah, was instrumental in gathering and collating the data cited in this study, and her expert help is gratefully acknowledged. The full data set on which this discussion is based can be seen here; for a manipulable copy of the original spreadsheet, please contact the author.
In late 2010, I was thinking quite a bit about book use in research libraries. The conventional wisdom was that “no one uses print books anymore” in libraries like mine, and indeed annual data provided by the Association of Research Libraries (ARL) showed a pretty clear decline in book circulations: between 1991 and 2008 (the most recent data available at that time), the number of initial circulations in ARL libraries had fallen by over a quarter. And when I ventured into the book stacks in my own library I usually found them spookily deserted.
But I was haunted by a passing comment a colleague had made to me a few years earlier, noting that the conclusions we draw from library usage data can easily be confounded by changes in the library’s user population. It occurred to me that if we really want to understand what’s happening with regard to library patrons and printed books, we need to take into account the changing nature of our patron base. And the simplest and most consistent change in that population is growth over time: university enrollment tends to grow from year to year.
The question I decided to examine, then, was: how much of the change in individual patron behavior is being hidden by raw circulation data? Clearly, if the size of your patron base is growing while circulation numbers remain the same, that means that the average patron is using the printed collection less; and if the circulation numbers are actually falling while your patron base is growing, that means the average patron is using the library at a more steeply-declining rate than the circulation data suggest.
The way to answer this question is simple, but not easy: you track circulation data for a library along with changes to the size of the patron population, dividing the number of circs by the number of patrons, and watch for trends over time. It’s not a particularly challenging statistical task, but gathering the data for a large number of libraries is a lot of work.
So for several months I spent part of each day gathering circulation and enrollment data for each of the 114 ARL member libraries from the ARL Statistics database. I examined the years 1995-2008, calculating two figures for that period for each library: the total change in raw circulation (number of initial circs), and the change in circulation rate (number of circs per full-time student). I counted only initial circulations rather than total circulations (in order to exclude renewal transactions from the calculation, since the point of the study was to measure how many times the books were used, rather than how intensively). I used enrollment as a proxy for “patron base,” even though I recognize that faculty, staff, and community patrons check out books from the library as well. Since these other populations remain relatively constant in size, and since I was more interested in tracking the shape of the change curve than in knowing its exact distance above the X axis, this seemed like a good compromise.
What I found was sobering: on an aggregate basis, ARL libraries had seen a fairly steady number of initial circulation transactions between 1995 and 2008, with totals hovering between 36 and 40 million. However, during that same period the aggregate circulation rate fell by almost 50% (Figure 1)—a significant change in patron behavior, and one entirely masked by the raw circulation trend.
Perhaps just as interesting as the aggregate numbers was the variation in results by institution. While the great majority of ARL libraries had seen declines in both circulation numbers and circulation rates during the 1995-2008, in some cases the difference between raw number and rate was very dramatic and in others it was less so — and a handful of libraries had actually seen an increase in both.
What came next was, for me, even more interesting: I had a great deal of trouble getting my study published. I submitted it to a journal with a particular interest in collection development in research libraries, and it was rejected. Then I submitted it to a journal focused on library management; no luck there either. I eventually decided to submit it in an abridged version to a less formal venue, and the report was published in Library Journal under the title “Print on the Margins: Circulation Trends in Major Research Libraries.” The data set was too large to embed into the article, so I pulled out a few noteworthy institutional examples and provided a link to the full data set for those wanting to investigate further.
That piece was published in mid-2011, and over the past couple of years I’ve been feeling a growing sense that it was time to revisit the ARL data and see what’s happened in research libraries since. So, with the able and gracious assistance of my colleague Lydia Porter, I gathered and analyzed the circulation and enrollment data for the years 2008-2015 (the most recent data currently available).
The results (Figure 2) are very interesting. A couple of noteworthy things have happened since the last time I looked at the data:
- In 2009, what had been a fairly steady state in initial circulations between 1995 changed dramatically, and there has been a hard and steady decline ever since. Between 2009 and 2015, total initial circulations in ARL libraries fell by almost half (from 36 million to 19 million).
- During that same period, the decline in circulations per student has continued as well, though the rate of decline has slowed. In fact, the average number of initial circulations per student has stayed the same (at 7) for the past three years, suggesting that the decline may have bottomed out.
What does all of this mean? I would suggest the following take-aways:
- Aggregate data are interesting, but for any individual library, what matters most is the trend at that particular institution. Interested readers from ARL libraries are encouraged to consult the data set and see what the trend lines look like for their own organizations.
- As I pointed out in my original article, it’s important not to assume that declining circulation rates mean declining use of library resources overall, or decreasing engagement with library services. For example, a library that offers an increasing proportion of its information resources online should naturally expect to see a decline in the circulation of printed books — especially if there’s significant overlap in content across formats.
- That being said, any library that is seeing a steep decline in the use of its print collection should probably let that trend inform a serious examination of its space and budget allocations. (And given that the average decline in circulations per student since 1995 has been so dramatic — from 25 to 7, a 72% decrease — there are lots of research libraries seeing steep declines.)
- For book publishers, these numbers have significance to the degree that print book sales to libraries matter to them. Those sales may matter less today than they once did, but even so, it’s probably worthwhile for scholarly book publishers to keep an eye on this trend.
In closing, I want to acknowledge some of the obvious limitations of this data set: it reflects trends only in large North American research libraries, and will have limited relevance (for example) to college libraries in North America and to research libraries in other global regions. Also, by measuring initial circulations only, this study ignores changing patterns in in-house use of books. (Other libraries’ mileage will certainly vary, but at my institution the reshelving of books has declined even more precipitously than circulations, leading me to believe that in-house use of books is falling as well.) My hope is that the data provided here will prompt further examination of relevant data in other areas.