Last month I posted a piece here in the Kitchen titled “How Important Are Library Sales to the University Press? One Case Study.” That posting reported on a study that I did in collaboration with Dean Blobaum of the University of Chicago Press (UCP): using sales data from UCP’s 2012 imprints and library holdings data as recorded in WorldCat, we tried to get a sense of what proportion of those sales were represented by library purchases.

wychwood branch library
Wychwood Branch Library. (Source: Wikimedia Commons)

Today’s posting reports on a case study that looks at university press books from the opposite perspective: that of one library that buys them. While “importance” is a broad and vague term and I have no illusions about the ability of library data (let alone a single library’s data) to establish it in any comprehensive or fundamental way, there are facets of importance that can be established—or at least substantially hinted at—by library data, so I offer the following data and analysis for what they’re worth.

Value and Importance, Circulation and Usage

One facet of “importance,” and a very significant one to libraries, is usage. When deciding whether to maintain a journal or database subscription, one of the questions we most commonly ask ourselves is “How much is this resource being used?” Since a journal or database represents an ongoing cost and offers a growing collection of discrete documents, we tend to use cost per download as a rough-cut measure of the value of journal and database subscriptions. At my library, cost per download is not a strictly determining factor when it comes to cancellation decisions—but it is a strongly suggestive factor, and as it rises it leads us to look more closely at other factors.

When it comes to printed books, which represent (on a unit basis) a small up-front investment followed by an open-ended trickle of cost for storage and maintenance, we haven’t historically worried too much about measuring value—once a book is on the shelf and its primary cost has been sunk, there hasn’t been much incentive to invest time and energy in figuring out how valuable it is. But as space gets tighter in our libraries and circulation rates fall, it is becoming more important to examine the infrastructural investments represented by our printed book collections, and one way of doing that is by measuring usage.

The simplest and most obvious way of measuring usage is by counting circulations, or check-outs. To be very clear, usage is not a perfect metric of value and circulation is not a perfect metric of usage—however, it’s important to bear in mind that for a library, the “value” of a book, in a pure and abstract sense, is not the main driver of the purchase decision. What matters is not so much the book’s absolute value in the world (we don’t buy a book just because it’s “good”) as the book’s appropriateness and usefulness in the collection—or, in other words, its value to the population served by the library. As for the limitations of circulation data as a usage metric, those limitations are real. But circulation does reflect one very important aspect of usage.

Generating the Circulation Data

The datasets under examination in this study were drawn from three years of circulation records stored in my library’s integrated system (we use Aleph, a product of Ex Libris). Gathering and cleaning the data proved to be an onerous task, and I am greatly indebted to Ian Godfrey (our Head of Facilities, Collections, and ILS Management) for his patient diligence in running the reports, helping me refine the queries, and running them again, in some cases repeatedly as we kept turning up different problems with the data. The resulting datasets are freely available for download here, and include seven spreadsheets:

  • title lists showing all books that circulated in 2011, 2012, and 2013;
  • lists of all university press books that circulated during each of those three years;
  • a list of all university press books in the library collection as of July 2014.

The big challenge here was culling university press titles from among the other books in our catalog. We accomplished this by pulling records in which any of the following four terms appear in the 260 field of the MARC record (the field that contains publication data, including publisher name): “university of,” “university press,” “MIT press,” and “teachers college.” The resulting list is not entirely free of anomalies, but I believe it’s clean enough for our purposes. By making the data sets publicly available, I’m inviting others to arrive at their own conclusions in that regard, as well as to reuse the data sets for any other studies that they might make possible.

One anomaly in the data is worth noting up front, however. Since ebooks (which do not circulate) showed up as duplicate titles in the total UP title list and accounted for 8.2% of those titles, I reduced the UP Titles in Collection figure by 8.2% (or 20,680). The fact that none of those titles could have circulated meant that I had to eliminate that number of titles from the set before I could use it to calculate the percentage of UP titles that had circulated. So I randomized the UP title list, cut 20,680 titles from the bottom of the randomized list, and calculated the circulating percentage from the remainder. The total number of titles (UP and trade) available to circulate was calculated by a different process from the one that generated the UP titles available to circulate, so that larger list did not have the same duplication issue. However, it did include many circulating items that are not books (computer cords, iPads, headphones, a human skeleton, etc.), and I did my best to remove all such irrelevant entries. A few certainly remain. I considered removing high-circulating textbooks, some of which we keep on course reserve, but decided that removing them would not be justified. A book is a book, and a checkout is a checkout.


Here I will present and briefly discuss some of the findings that I’ve gathered from this circulation data.

First, the general circulation data illustrates a well-established and ongoing trend in large research libraries: the declining circulation of printed books (see Table 1). It’s interesting to note, however, that of the books that circulated at all, the percentage of them that circulated more than once has been growing. This suggests that while fewer and fewer books are being used, those that do get used are being used more heavily.

Table 1


Total Books Circ’d

% of Titles Circ’d >1










Second, university press titles represent about 9% of total printed books in the library collection. It’s important to note that this reflects current holdings; it was not possible to measure this percentage in previous years.

Third, university press titles account for about 17.5% of book circulations annually (see Table 2). This is interesting: although UP titles represent quite a small minority of items in the print collection, they punch well above their weight in terms of usage. On the other hand, given that ours is a large research library, it should probably be expected that more scholarly books would get heavier use.

Fourth, of university press titles that circulate at least once, the percentage that circulates more than once is falling (see Table 2). This is also an interesting and potentially troubling finding: this metric has seen a drop of 8.4% over the three years studied.

Table 2


% of Circs That Are UPs % of UP Titles Circ’d >1







2013 17.5



  1. Obviously, circulation is an imperfect representation of book usage; some important book usage occurs in the library and never leads to a checkout. Or used to, anyway: in my library, in-house usage (as reflected by the reshelving of books by library staff) has been declining even more rapidly than circulation. In 2011, our staff reshelved 248,428 books; in 2013, they reshelved 179,663, representing a 28% decline in only two years.
  2. The data in these sets is not perfectly clean; given their size and the fact that library systems were never intended to provide this kind of data, that’s inevitable. However, I’m pretty confident that the data is good enough to go on. If anyone finds anomalies that cast serious doubt on the findings above, I welcome the input.
  3. Over the past five years, a growing proportion of our book purchases have been ebooks, which certainly has had some impact on circulation of university press (along with all other) books.
  4. This is a case study from one particular library, not a comprehensive study of research libraries generally, still less of academic libraries as a whole. These findings are, at best, only imperfectly generalizable to the larger population of academic libraries. My hope is that these snapshots will be helpful and will prompt others to undertake similar studies.
  5. The questions that led me to undertake this study were actually about scholarly monographs in particular, not university press books in general. So that’s the next step I hope to take. The challenge, though, will lie in figuring out how to identify a large-enough sample of UP titles from these lists that unambiguously fit the profile of a scholarly monograph.
Rick Anderson

Rick Anderson

Rick Anderson is University Librarian at Brigham Young University. He has worked previously as a bibliographer for YBP, Inc., as Head Acquisitions Librarian for the University of North Carolina, Greensboro, as Director of Resource Acquisition at the University of Nevada, Reno, and as Associate Dean for Collections & Scholarly Communication at the University of Utah.


45 Thoughts on "How Important Are University Press Books to the Library? One Case Study"

An idle thought – which you allude to in caveat #5 – if you scan the top #100 well-used UP items, do very many scream “these aren’t scholarly monographs”?

I’m not very familiar with the US university presses, but certainly this would be plausible in the UK – lots of OUP editions of core literary texts, for example.

Hi, Andrew —

That’s a good thought. One problem, though, is that scanning the top 100 circulators won’t really tell us much, since all you can see in the report is the first element of the title. And as we all know, scholarly monographs often give themselves away only in the subtitle (“Living la Vida Loca: The Insanity Defense in Rural Peruvian Agricultural Litigation, 1750-1759”). But it may well be worth the effort to drill down into those titles further to see what patterns emerge.

The moral of the comments below is clearly that someone should publish this particular title ASAP and cash in on the demand…

I agree that there’ll be some false positives and a few crosschecks needed, but it should hopefully be easy to spot the most obvious things like editions of Herodotus or Swift from the author fields, or material like the Cambridge Collections.

Well, the data sets are publicly posted now (you’ll find a link in the posting above) so feel free to take a look and see what you think.


I’ve skimmed every “university” title with >5 edits and it looks like there’s very little if any of this kind of material – a few textbooks, certainly, but little in the way of literary texts etc. So scratch that hypothesis…

Interesting data. But how do you now answer the question posed in the headline to your post?

As I explained in the second paragraph, the data I’ve been able to gather don’t permit me to answer the question in any kind of complete or categorical way. The data available can provide hints about certain kinds of importance in the context of one research library. My hope is that other libraries will conduct similar studies, and that more generalizable patterns might emerge as more studies are done.

Ok, then let me ask the question this way: if it were found that every university library similar to yours had the same numbers, how would you answer the question you have posed?

In that case, I guess I’d have to say that the answer is complicated. On the one hand, UP books account for a higher number of circulations than their representation in the collection would suggest. On the other hand, the number of uses that a typical UP book gets in a year is declining. And then there’s the fact that circulations of all books (including UP titles) is declining every year. So there doesn’t seem to be a simple answer to the question I’ve posed. And the reality, of course, is that in the real world the answer will be not only complicated, but also different from institution to institution.

It strikes me as a problem that you categorize books in university libraries as either “university press” or “trade”; what about the many scholarly monographs that are published by houses such as Polity, Routledge, and yes, Lynne Rienner Publishers. I can’t help but wonder what your usage statistics would look like if all peer-reviewed scholarly monographs were included in the mix.

A few trade houses do indeed publish scholarly monographs on a regular basis. But this particular study is focused specifically on the distinction between UP books and non-UP books, not between scholarly monographs and non-scholarly books. As I mentioned in the posting, I would like to be able to examine the latter distinction, but one major stumbling block is the fact that there’s no systematic way of discriminating between scholarly monographs and other books.

Your search criteria would result in including the University Press of America–a quasi-vanity press–and omitting the Brookings Institution Press (a full member of the AAUP). Did you correct for such anomalies?

I have no problem with measuring the “importance” and “value” of books in this way so long as it is understood in the sense you use here, which does not include the meanings these terms have in their “pure and abstract sense.” The latter would be better measured by such criteria as awards given by scholarly associations.

Your search criteria would result in including the University Press of America–a quasi-vanity press–and omitting the Brookings Institution Press (a full member of the AAUP). Did you correct for such anomalies?

Nope. As for AUP: that’s a good point, but we in libraries tend to be quite aware of its status and don’t buy much from them. In the case of the data sets under consideration here: in 2011, of the 22,881 UP titles that circulated, two were from AUP; in 2012, of the 21,121 UP titles that circulated, three were from AUP; in 2013, of the 17,949 titles that circulated, none was from AUP. So failing to filter them out did not result in any meaningful distortion of the data.

As for the Brookings Institution Press: that’s a tough one. Yes, BIP is a member of AAUP. But does that make it a university press? The Brookings Institution is a think tank, not an academic institution. (You could join the American Library Association if you wish; that wouldn’t make you a librarian.) The AAUP’s attempt at defining a university press leaves, in my opinion, quite a bit to be desired. And given its rather loose membership criteria (“The American Association of University Presses is a membership organization of nonprofit scholarly publishers”), I don’t think AAUP membership is a good inclusion criterion for a study like the one I’m reporting on here.

> And given its rather loose membership criteria …

Ouch, Rick … did you really mean to say that? Like that? Although I’ve only been in the saddle for 18 months, I’ve consistently heard exactly the opposite; i.e., our membership criteria are too rigid, and why can’t we relax them. If you click on the “Guidelines for Membership” and “Eligibility Requirements” (on the same page you link to) you’ll get just a little flavor for what’s required to join. And even that’s not the whole story … I’m told by the press directors who’ve gone through it that our application review process is extensive – even intrusive — and exhaustive. AAUP has various membership categories, and I’d agree that certain of these might not suit your purpose, but that’s not quite the same as saying our criteria are “loose”. There are, after all, only 134 of us!

Peter, apologies if my comment came across as critical. Let me clarify: when I characterized AAUP’s membership criteria as “loose,” I meant that only in the sense provided by the context of Sandy’s comment. In other words, Sandy was suggesting that the Brookings Institution Press ought to be considered a “university press” because it’s an AAUP member. But it’s clear that a press does not have to be a university press to be a member of AAUP, because AAUP defines its potential membership more “loosely” than that. That does not (in my view) reflect in any way negatively on AAUP as an organization or on the quality of the presses it includes as members; it only reflects negatively on the argument that AAUP membership qualifies a publisher to be characterized as a “university press” for the purposes of a study like mine. Does that make sense?

That said, I stand by my comment that AAUP’s definition of “university press” leaves much to be desired. But that doesn’t reflect negatively on the organization either.

Thanks for the context, Rick. And I agree the “definition” certainly deserves a fresh look – thanks, too, for bringing that to my attention!

Reblogged this on and commented:
Last month, Rick Anderson at the scholarly kitchen asked what percentage of university presses book sales were to libraries. This month he has turned the question around and asked what percentage of books circulated by libraries were published by university presses. Anderson has some interesting findings about the of university press books to libraries.

Rick, very interesting preliminary study. I do understand why you focus on the print publications of university presses. But our e-book collections of university press titles is very quickly outpacing in numbers and use the historical and current holdings in print. So we need to find a way to combine both print circulation and consultation of digital holdings into the analysis in order to get a good picture of use going forward. I note that as the total books circulation has gone down, from 130k to 103K, the circulation of university press books has similarly declined from 23k to 18k. I wonder how much of this decline can be explained by the expanding availability of digital versions.

Jim, that’s a good question. The circulation decline in large research libraries generally predates the wide availability of ebooks, but that doesn’t mean ebook availability hasn’t played an important role in the ongoing trend (I suggested that connection in Caveat #3, above).

And I agree that if we really want to understand usage patterns for university press content generally, we’ll need to find a way to look at both ebook and print-book usage–that would be a good topic for another study.

This may be off topic but you describe the cost per download as an important, albeit suggestive, criterion for continuing journal subscriptions. Suppose a journal’s articles become free after say twelve months, because of a green open access mandate. Would that change the cost per download calculation, given that the cost might be figured now as only for the first twelve months of downloads, the rest being free?

To answer your question, yes: it would greatly increase the calculated cost-per-download and make it less likely that we’d continue our subscription.

If UP books make up only 9% of your total printed-book collection but “punch well above their weight” by accounting for 17.5% of circulation, why doesn’t your library–guided as it it by “value to the population served by the library”–buy more UP books and fewer of that other 91%?

Well, bear in mind that there aren’t nearly as many UP titles available for purchase in any given year as there are non-UP titles, so any large research library collection is inevitably going to consist mostly of non-UP books. It’s also true that as we continue moving more and more in the direction of patron-driven acquisition, this question will increasingly be a moot one.

But it does seem to suggest that U. presses are not publishing enough books, which has been my contention for some time. My guess is that your library, which is large but not among the largest, already purchases all the U. press monographs, but skips on the U. press trade titles. But that’s another bit of research.

Agreed. There are so many possible directions for fruitful research here.

Rick, one of my concerns regarding usage as a measure of value is that the nature of the university, in the best sense of the word, is to cultivate study broadly and deeply. It should not be motivated by economic, political or popular interests. As this argument over the value and quality of UP titles has continued over the past month or two, I have had occasion to review some UP new title lists in detail (a professional hazard). I have kept your arguments in mind recently while reviewing these lists and wondered if how much usage there could possibly be for many titles. And so would a lack of usage diminish their value to the scholarly enterprise? The question I’m struggling with is what is the future of the library? If not to gather these often rare pieces, to build collections in an intelligent way for both discovery and preservation, then what?…. The highly used materials are probably the ones Google can make discoverable and available more easily (and economically) than libraries. I think usage is a metric that we need to be very careful of – and be aware that usage of journals is a different thing than the usage of books (there is a fundamental difference between purposes of the formats for one thing).
Anyway, your work is important and I think we all appreciate your leadership in taking it on. It gives the rest of us something to talk about!

Hi, Mike —

Thanks for your (as always) thoughtful comments. You raise important issues, and I think that in responding to them I need to repeat and emphasize a couple of things I said in the posting:

1. There’s a big difference between measuring the value of a book in universal or absolute terms (let’s call that “Value” with a capital “V”) and measuring the value of a book to a particular library or collection (let’s call that “value”). Circulation data can’t tell you whether or not a book is Valuable, but it can give you an important clue as to its value to the library that bought and cares for it.

2. You object to the idea of using usage as a measure of Value, because the university’s mission and decisions “should not be motivated by economic, political, or popular interests.” But there are a couple of problems with that statement. First, while you and I might agree that the university shouldn’t be motivated by financial interests, the fact remains that the university’s options are limited by financial constraints. This is why libraries have to make difficult purchasing decisions, and it’s why those decisions can’t be based purely on quality or on Value. No library can afford to buy, house, and care for all of the Valuable books and journals that exist, so every library has to choose from among those Valuable books and journals based on some additional set of criteria.

3. This brings us to the other problematic element of your statement: your objection to making choices based on “popular interests.” In the context of a research library, what you’re calling “popular interest” (as expressed in usage) can more accurately be called “scholarly use” — since the users who have access to the library are, for the most part, students and faculty. So the term “popularity” is, in this context, a bit tendentious and misleading.

4. It’s also true that the academic library’s job is not simply to acquire as much Valuable content as it can. Its job is to support the academic work of its host institution. That means that it may also need to make space in its budget for books that are less Valuable than some others, if those books are needed to support the specific programs and disciplines on which that institution focuses its work. As I tell vendors on a regular basis: if my university has a high-energy physics program but no architecture program, then our students and faculty may be better served by a pretty-good book on high-energy physics than by the world’s best book on architecture. In other words, it’s not all about Value — it’s also about relevance and utility. Usage isn’t a perfect measure of utility, but at the local level (where the purchasing decisions have to be made) it’s arguably a better measure than Value is, strictly speaking.

5. I disagree that the library’s purpose is to “to build collections in an intelligent way for both discovery and preservation.” I would argue that collections are a tool, not a goal. We build collections so that people will have access to what they need in order to do their scholarly work, not so that we will have wonderful collections. In the current information environment, new possibilities are emerging that allow us to provide access by means of processes that don’t necessarily result in the building of collections (as traditionally understood). To reject those opportunities because they don’t result in collections, even if they serve students and scholars well, is to confuse ends with means.

I have no disagreement with your reply, but I think the problem Mike is struggling with can best be resolved by divorcing Value from the market altogether. If we publish Valuable books just because they are Valuable and not necessarily because anyone wants them for immediate pragmatic use, the best model to use is OA, funded ideally by endowment money. This is what Amherst College Press will be doing, and we can hope that others will eventually go down that same path. The market as it exists now is a definite constraint on the publication of Valuable books if not books of value.

I think the problem Mike is struggling with can best be resolved by divorcing Value from the market altogether.

That may be the case. And when the great day arrives that Valuable books cost nothing to acquire, and can be stored and curated without cost, then there will be no reason for any library not to acquire all of them. Of course, if the library can acquire these books without cost and doesn’t have the responsibility of storing and curating them (which is the only way we could do those things without cost), then in what meaningful sense does providing access to them mean “acquiring” them? To the degree that doing those things continues to mean spending money or allocating space, the need to make difficult decisions about limited resources will remain.

Thanks for your response. I’m slow in responding – I was fine until suddenly, after reading “in the current information environment” (in point 5), I decided I needed a vacation 😉 We agree on your points 1-4 entirely (I think we just misunderstood – not disagreed – on the meaning of “popular” – I of course meant it in the context of the university and not in the sense of ‘public library’ or other areas). I enjoyed your distinction of ‘Value’ and ‘value’, as well as the comment on ‘useful’ and ‘usage’. But point 5 is problematic. I understand and appreciate the distinction between the ‘useful’ and the ‘good’, but when we arrive at the conclusion that “in the current information environment, new possibilities are emerging…is to confuse ends with means” to reject collection building is a bit of sophistry. No one would argue that the means of building a collection aren’t changing and that the library must be a pile of books in a particular building. The argument regards your use of usage as a measure of utility (in the article above) and the contention that since many university press books have little value by this measure (also in recent posts on Liblicense), should be ‘collected’ using “new possibilities”. We all know what usage rates are for the scholarly monograph. For a scholarly monograph (regardless of format, e or p) there may well just be a single use in the first 3 years of its life in a library, so usage is going to be a poor measure of utility (you yourself suggest that in point 4). Does this diminish its ‘value’ if not its ‘Value’? The real issue underlying the ‘new possibilities’ comment and the ‘collections’ criticism is the whiff of the idea that information is getting freer and cheaper (as supported by the 2 comments following our original exchange – you and Sandy Thatcher). The role of the academic library, the stresses and forces redefining the nature of the academic library (a collection, a resource, a tool?), are affecting the entire ecosystem: publishers, vendors, service providers, etc. While information is apparently getting freer and cheaper for libraries, unfortunately it is not yet getting freer and cheaper for the other parts of the ecosystem. And with this, don;t we throw open wide the doors of the argument 🙂
More than the lunches will be lively with this topic at Charleston in November.

PS. As an aside, there was once an argument that ‘Approval Plans’ would result in ‘vanilla collections’ across the academic library world. I wonder if ‘usage’ will be impugned in the same way?

For a scholarly monograph (regardless of format, e or p) there may well just be a single use in the first 3 years of its life in a library, so usage is going to be a poor measure of utility (you yourself suggest that in point 4). Does this diminish its ‘value’ if not its ‘Value’?

I don’t think the question is whether lack of use diminishes a book’s value; the question is whether the lack of use tells us anything about its value. I think it does — again, bearing in mind my distinction between “value” and “Value.” Can we say, categorically, that a library book that has sat, untouched, on the shelf for five years has zero value? Well, we can say that it has produced no value for that library’s patrons yet — but we can’t say that it will never produce any value. Of course (as I’ve pointed out here before) the problem we face in libraries is that we don’t have the luxury of making acquisition decisions based purely on potential value, because the number of books that have potential valuable is functionally unlimited and our resources are very strictly limited. This is one of the major problems with traditional, speculative collection-building: it’s a process by which we allocate strictly limited resources in the building of resources the value of which has not been demonstrated, but is only hoped for. In other words, the problem we face isn’t determining which books have “Value”; the problem is, given severely limited budgets, how to select from among them the ones that will also have significant “value.”

While information is apparently getting freer and cheaper for libraries, unfortunately it is not yet getting freer and cheaper for the other parts of the ecosystem.

Actually, I disagree with this premise. Nothing that libraries buy on behalf of users is getting cheaper. But an awful lot of the information to which library users have access outside of the library — much of it information that used to be scarce and expensive — is indeed getting freer and cheaper. That puts pressure not only on publishers and vendors that need revenue streams in order to stay in business, but also on libraries, much of whose value proposition in the past lay in the brokerage function. Our patrons simply don’t need us in the same ways they used to. We run the risk of going out of business too (though we don’t like to admit that).

As an aside, there was once an argument that ‘Approval Plans’ would result in ‘vanilla collections’ across the academic library world. I wonder if ‘usage’ will be impugned in the same way?

Maybe it will, and when it does I’ll keep responding the same way: I don’t think we should worry too much about whether or not our collections are “vanilla.” What we should worry about is the degree to which our collections help or hinder the scholarly work of our students and faculty.

Just curious, Rick, how do you know that a book has been “untouched”? Does your library follow a rule that all patrons must NOT return books to the stacks that they take off the shelves, so that library staff can count how many times a book was taken of the shelf and presumably “used” in some manner? Or is yours a closed-stack library so that no patron can get any book unless it is “checked out” in some way, even if never taken off the library premises?

In my library (as in most academic libraries, in my experience) we ask patrons not to reshelve their books — partly because of the risk of misshelving, and partly because we want to track reshelving patterns. This means, of course, that in no real-life case can we ever say for certain that a book has been absolutely untouched. But I was using the concept of the absolutely untouched book as a theoretical framing device; in real life, the application of the principle it illustrates is relative: “Do I have good reason to believe books in category X are getting usage sufficient to justify the expense of acquiring and keeping them?”; “Given the purposes of our collection, would it make more sense only to acquire books in this category where demand has been demonstrated in the real world?”; etc.

Very interesting stuff. Thanks for sharing. I find it particularly interesting, to begin with, that university press publications represent such a small percentage of the books on the shelves, since they represent a rather large percentage of things I have read, and things I have cited.

Comments are closed.