4
Editorial Some Thoughts on Academic Library Collections by Donald W. King Available online O ver the past four years, colleagues and I have con- ducted a series of studies to examine the impact of electronic journals on the use and cost of academic collections. In particular, Carol Tenopir, University of Tennes- see, Carol Hansen Montgomery, Drexel University, and staff at the University of Pittsburgh (Sarah Aerni, Fern Brody, Matt Herbison, and Amy Knapp) and I surveyed faculty and students to determine their overall information seeking and reading patterns. Detailed cost studies were performed at Drexel Uni- versity (Carol Hansen Montgomery), University of Pittsburgh (staff including Paul Kohberger), and nine other academic libraries: Bryn Mawr, Cornell, Franklin and Marshall, George Mason, NYU, Suffolk, Western Carolina, Williams, and Yale (Roger Schonfeld, Ithaka; Eileen Gifford Fenton, Ithaka; and Ann Okerson, Yale University). These studies provide evidence to address contemporary concerns of academic librarians and university administrators. ARE SCHOLARLY JOURNAL COLLECTIONS WORTH CONTINUING? Evidence not only suggests that scholarly journal collections are worth continuing, but they appear to be even more used and useful than twenty-five years ago. Faculty at Tennessee, Drexel, and Pittsburgh universities average 206 annual readings of scholarly articles and spend an average of 126 hours reading these articles. Clearly faculty would not expend their most important resource, their time, on reading articles if they did not achieve corresponding value from the information read. In fact, about one-half of reading is done for primary or background research, one-fifth for teaching, and the remainder for writing, keeping up with the literature, and so on. The information read is rated highly important to research and teaching and over half of these readings are said to improve research or teaching. Faculty who had recently received awards or special recognition read one-third more than others, and the extent of publishing is correlated with amount of reading. These indicators are all found to be more favorable with reading from library collections. Faculty were asked, for their last library reading, what they would do to get the information if there were no library at their university. Taking into account instances in which readers would not do anything, the readers indicate they would have to spend additional time and money to obtain the articles, which totally would cost the university about double the current journal collection costs. I did a series of readership studies in the 1970s and early 1980s under contract to the National Science Foundation and ever since have kept track of science readership in forty other readership studies done over a twenty-five-year span. In 1994, Carol Tenopir and I began extracting survey responses of university (and other) scientists from national surveys (1977 and 1984) and surveys done in universities (1993, 2000, 2002, and 2003). From 2000 to 2003, university scientists (including engineers and social scientists) were found to average 216 annual readings versus 175 readings for the other faculty. We also observed a steady increase in amount of reading over time; in fact a 44 percent increase in reading by university scientists over twenty-five years. What is remarkable is that the amount of increase in reading is almost exactly the same as the increase observed from academic library collections over the twenty-five-year period (sixty-six versus sixty-four additional readings per year). Thus, the in- creased reading appears to come from the library collections. There are several plausible reasons for this phenomenon. First, the number of personal subscriptions has decreased from 4.2 to School of Information Sciences, University of Pittsburgh, 135 North Bellfield Avenue, Pittsburgh, PA 15260, USA The Journal of Academic Librarianship, Volume 30, Number 4, pages 261–264 July 2004 261

Some Thoughts on Academic Library Collections

Embed Size (px)

Citation preview

Editorial

School of In

The Journal of

fo

A

Some Thoughts on Academic LibraryCollections

rmation ScienBellfi

cademic Librar

by Donald W. King

Available online

Over the past four years, colleagues and I have con-ducted a series of studies to examine the impact ofelectronic journals on the use and cost of academic

collections. In particular, Carol Tenopir, University of Tennes-see, Carol Hansen Montgomery, Drexel University, and staff atthe University of Pittsburgh (Sarah Aerni, Fern Brody, MattHerbison, and Amy Knapp) and I surveyed faculty and studentsto determine their overall information seeking and readingpatterns. Detailed cost studies were performed at Drexel Uni-versity (Carol Hansen Montgomery), University of Pittsburgh(staff including Paul Kohberger), and nine other academiclibraries: Bryn Mawr, Cornell, Franklin and Marshall, GeorgeMason, NYU, Suffolk, Western Carolina, Williams, and Yale(Roger Schonfeld, Ithaka; Eileen Gifford Fenton, Ithaka; andAnn Okerson, Yale University). These studies provide evidenceto address contemporary concerns of academic librarians anduniversity administrators.

ARE SCHOLARLY JOURNAL COLLECTIONS

WORTH CONTINUING?

Evidence not only suggests that scholarly journal collectionsare worth continuing, but they appear to be even more used anduseful than twenty-five years ago. Faculty at Tennessee, Drexel,and Pittsburgh universities average 206 annual readings ofscholarly articles and spend an average of 126 hours readingthese articles. Clearly faculty would not expend their mostimportant resource, their time, on reading articles if they didnot achieve corresponding value from the information read. Infact, about one-half of reading is done for primary or backgroundresearch, one-fifth for teaching, and the remainder for writing,

ces, University of Pittsburgh, 135 Northeld Avenue, Pittsburgh, PA 15260, USA

ianship, Volume 30, Number 4, pages 261–264

keeping up with the literature, and so on. The information read israted highly important to research and teaching and over half ofthese readings are said to improve research or teaching. Facultywho had recently received awards or special recognition readone-third more than others, and the extent of publishing iscorrelated with amount of reading. These indicators are all foundto be more favorable with reading from library collections.Faculty were asked, for their last library reading, what theywould do to get the information if there were no library at theiruniversity. Taking into account instances in which readers wouldnot do anything, the readers indicate they would have to spendadditional time and money to obtain the articles, which totallywould cost the university about double the current journalcollection costs.

I did a series of readership studies in the 1970s and early1980s under contract to the National Science Foundation andever since have kept track of science readership in forty otherreadership studies done over a twenty-five-year span. In 1994,Carol Tenopir and I began extracting survey responses ofuniversity (and other) scientists from national surveys (1977and 1984) and surveys done in universities (1993, 2000, 2002,and 2003). From 2000 to 2003, university scientists (includingengineers and social scientists) were found to average 216 annualreadings versus 175 readings for the other faculty. We alsoobserved a steady increase in amount of reading over time; infact a 44 percent increase in reading by university scientists overtwenty-five years.

What is remarkable is that the amount of increase in reading isalmost exactly the same as the increase observed from academiclibrary collections over the twenty-five-year period (sixty-sixversus sixty-four additional readings per year). Thus, the in-creased reading appears to come from the library collections.There are several plausible reasons for this phenomenon. First,the number of personal subscriptions has decreased from 4.2 to

July 2004 261

3.5 subscriptions per scientist, with resultant decreased readingsfrom personal subscriptions likely absorbed by reading fromlibrary collections. Second, much more reading now comes fromonline bibliographic searches, and along with articles identifiedfrom citations, many articles are identified that must be locatedand obtained before they can be read and libraries are obviouslythe most reliable and trusted source for obtaining these articles.The third reason library collections are being used more is recentaccess to enlarged electronic collections. Clearly, scholarlyjournal collections should be continued!

ARE ELECTRONIC JOURNAL COLLECTIONS

SUFFICIENTLY USED?

Evidence from our recent surveys involving both faculty andstudents (Drexel 2002 and Pittsburgh 2003) suggests that theelectronic journal collections are extensively used. Here weobserve readings as a use metric rather than publisher- andvendor-provided ‘‘hits’’ because of inconsistencies with the latterdata. While nearly all reading from personal subscriptionscontinues to be from print issues (even though most subscrip-tions are available in electronic format), about 80 percent offaculty reading from the library collections is from electronicversions accessed mostly from their office, lab, home, etc. Oneobvious reason that faculty prefer to get their library articlesremotely is that it saves faculty about twenty hours per year perperson, which can be allocated to more useful purposes. On theother hand, this means that photocopying and online printout areshifted from the library to departments and other locations. Mostreading of library collections is by students (about 80 percent ofall the reading), although less of their reading is from theelectronic collection (71 percent) and more is done in the mainlibrary or their department library.

To place a context on this amount of reading, at Pittsburgh,2500 faculty and staff and 30,100 students had 1.3 millionreadings from the library collections, with 930,000 of thesereadings from the electronic collection. The 2003 sum ofpublisher and vendor ‘‘hits’’ is 1,320,000 or 42 percent higherthan readings, and hits at Drexel were one-third higher than thesurvey estimate of amount of reading. Whether observed byamount of reading or online ‘‘hits,’’ evidence suggests thatelectronic collection use is substantial and, as mentionedabove, contributes to an increase in reading and reading fromlibrary collections. The reading from electronic versions maydominate academic collection use in most universities andcolleges.

HOW MUCH DO PRINT AND ELECTRONIC

COLLECTIONS ACTUALLY COST?

We examined collection costs from two perspectives: as anannual budget and the average cost of a print and an electronicjournal subscription over a twenty-five-year life cycle. In bothinstances, all studies of the eleven academic libraries suggest thatelectronic journals tend to cost less than the print versions. Here Iuse Pittsburgh’s cost analysis as an example because the printand electronic collections were about equal size at the time costwas estimated (i.e., 16,924 print subscriptions and 14,284electronic titles acquired). The total annual cost of these twocollections is US$8.2 million, including all resources (i.e.,purchase price, staff, space, shelving, workstations, equipment,systems, etc.) and with overhead allocated. About one-third ofthe total cost (US$8.2 million) is for the electronic collection and

262 The Journal of Academic Librarianship

two-thirds the print collection, even though the collection sizesare roughly equal. However, since the print collection includes alarge back file of bound volumes, perhaps a more usefulcomparison is the cost per reading, which is US$3 per readingfor the electronic collection versus US$15 for the print collec-tion. The twenty-five-year life cycle cost at Pittsburgh is aboutUS$180 per electronic title and US$580 per print title, againfavoring the electronic collection.

On the other hand, some electronic-related activities costmore than print or the costs are simply unknown, such as inarchiving. This is illustrated by examining staff time and cost.One set of activities involves initial processing of the collections(i.e., collection development, licensing and negotiations, acquis-itions, receipt processing, and cataloging). The staff is moreproductive when processing the electronic titles compared withprint subscriptions (1.2 electronic titles processed per hourversus 0.4 subscriptions per hour) and the unit cost is less(US$35 per title versus US$57 per subscription). The time andcost of some activities tend to vary by size of the user commu-nity (e.g., training sessions, user instruction, guides, facultyliaison, etc.). These activities tend to be more expensive forelectronic support than print. For example, the cost of training,briefings, and user instruction for electronic support is aboutUS$6.20 per participant versus US$4.70 per participant for printsupport. It appears that the cost of training, etc., may be morethan doubled due to adding the electronic collection. The librarystaff also spends substantial time supporting use of the twocollections, but comparatively more for the print collection. Use-related activities (i.e., reference and research, online searching,circulation, photocopying, reshelving) cost about US$0.60 perreading of electronic articles versus US$1.90 for print. There is atendency for staff hourly costs to be higher for activitiesinvolving electronic journals, which suggests that migrationtoward electronic collections may require shifting to a higherlevel of staff.

It is important to note, however, that virtually no staff or costsare currently attributed to activities related to the long-termpreservation of electronic journal collections. While there isclear agreement that some mechanism must be developed toensure the longevity of these important resources, a permanentarchiving solution has not yet emerged. Consequently, the relatedcosts are not yet known nor reflected in any analysis done todate.

The discussion above includes the cost of back files of boundvolumes. However, it is useful to distinguish current and backfile collections. At Pittsburgh, the back file collection represents45 percent of the entire journal collection budget (with the printpurchase price allocated in proportion to use). The unit cost perreading is US$15.60 compared with US$13.60 for currentperiodicals and US$3.00 for the electronic collection. However,the cost per reading from the back file is likely to rise sharply asthe electronic collection begins to replace the print collection andits use declines. For example, at Drexel, where the collection isnearly all electronic, the cost per reading of the back file isUS$30.00 per reading and it will continue to increase as usefrom this collection dwindles. The fact that JSTOR and manypublishers have converted back issues to electronic formatmeans that use from the electronic collection will rapidly replaceback file use. Even now at Pittsburgh, electronic journalsaccount for about two-thirds of reading of journals less thanfifteen years old and one-half of reading of journals over fifteenyears old.

I believe that academic libraries must reconsider their policiesconcerning back files in the future when there is reasonableassurance that electronic journal archives are permanent andaccess guaranteed and reliable. This does not mean that all printcollections should be discarded, but that cooperative collectionsand/or storage may be warranted in the future.

WHY DID LIBRARY SUBSCRIPTION PRICES

INCREASE SO MUCH?

There is no question that scholarly journal prices, particularlyin science and medicine, spiraled upward over the last twenty-year period during which print journals dominated. The commonreason given for this is that commercial publishers, especiallyElsevier, gained a monopolistic position in which market powerallowed them to charge high prices. However, there are otherfactors that generally contributed to the price increases as well. Ihave tracked the characteristics of a sample of science journalssince 1960 and the average price of these journals increased at arate higher than inflation nearly every year since then (beginningwith an average price of less than US$10 in 1960 and at a timewhen personal and library prices were about equal).

In addition to inflation, a major contribution to the priceincrease is that the average size of science journals has increasedby 80 percent over the last twenty-five years, thus requiring asimilar increase in price. While faculty personal subscriptionsdecreased modestly, average nonuniversity personal subscrip-tions decreased from 5.8 to about 2.5 subscriptions, whichaltogether resulted in about US$1 billion in lost revenue topublishers, some of which had to be recovered through anincrease in library subscription prices. Another factor is thattwenty-five years ago, many science society publishers requiredauthor fees (e.g., page charges, submission fees, reprint orpreprint fees, revision fees, etc.) and the revenue was used topay for some of the article processing costs. However, in the1980s, other publishers, including commercial publishers, com-peted for manuscripts by not charging fees so that the feepractice subsided and the societies had to increase prices tolibraries in order to obtain a sufficient revenue to break even.Finally, in labor-intensive services such as publishing, there tendto be diseconomies of scale due to overhead. It may be that thelarge portfolios of commercial publishers may require higherprices due to overhead, although they may be able to provideadditional features that others could not due to large R&Dbudgets that are available to them.

My best guess is that inflation and increased size of journalstogether account for about 50 to 60 percent of the price increases,the reduction in circulation about five to 15 percent, substitutionof revenue from author fees 10 to 20 percent, and other factorsthe remaining proportion.

With scientists (and perhaps other faculty) reading more fromlibrary collections (i.e., from an average of 37 to 101 libraryreadings per scientist) coupled with fewer science journals perscientist, the amount of reading per collection title may haveincreased as much as four- or fivefold over a twenty-five-yearperiod. The implication is that the subscription purchase priceper reading may not have increased much more than inflationaryincreases, particularly if student reading increased as much asfaculty reading. On the other hand, the increased library readingwill have cost the library as much as US$1.90 per reading morein staff use support time.

We know that commercial prices (which reflect their costsand profits) tend to be much higher than those of other publish-

ers. Yet we do not know whether their costs are higher thanthose of other types of publishers. Academic libraries, throughsubscriptions and licenses, pay for the cost of processingarticles, that is, the information content provided by publishers,and the cost of access to the information content. For printsubscriptions, the latter cost tends to be less than US$50 persubscription for printing and mailing (or currently aboutUS$1.50 per reading at Pittsburgh) and is negligible for elec-tronic access. The real issue then is the cost of informationcontent. The controversy of open access by author payment hasforced many publishers to reveal their article processing costs tocounter what Biomed Central and Public Library of Sciencecharge authors. It appears that commercial publisher articleprocessing costs (including profit) may be similar to those ofother publishers. The reason then that commercial publisherprices are higher than other publishers is due in part to therelative larger size of their journals and their circulation tends tobe much smaller, thereby requiring higher prices in order torecover the total cost of article processing.

ARE BIG DEALS GOOD DEALS?

This question is difficult to answer because many ‘‘big deals’’are negotiated and such negotiations may vary by size ofcollection involved, duration, and frankly negotiation ability.Both Drexel University and University of Pittsburgh appear tohave good deals. Evidence suggests that overall smaller librarieswill benefit more than large research libraries. There is, ofcourse, a concern of what will happen in the future as licensesare further negotiated. Libraries will have to consider whetherentire packages are worthwhile or subsets of them should benegotiated.

One characteristic of publishing that is not well known isthat a few highly read articles in a journal tend to subsidizeothers and some journals in a publisher’s portfolio of journalssubsidize others. While this may appear to be inefficient, itdoes guarantee that high quality articles and journals that havesmall, specialized audiences are published. The increased sizeof journals, mentioned earlier, is a trend in which the distribu-tion of reading and subsidization becomes more acute. In fact,some ‘‘mega’’ journals that have thousands of articles aresomewhat like ‘‘little deals’’ in which libraries purchase themeven though many articles in them may not be read by their users.The ‘‘big deal’’ packages are, in a sense, a further extension ofthis trend.

I think that academic librarians should consider use or betteryet the cost per use of their packages and, if cost per use is toohigh, portions of the packages considered for further negotiation.For decades, there have been studies showing that frequentlyused journals should be purchased and interlibrary borrowing ordocument delivery utilized for infrequently used journals. In fact,it is with this knowledge that librarians essentially finessed theinterlibrary loan issue in the 1976 revision of the Copyright Lawwith the fair use provision which stated that the borrowinglibrary is exempt from royalty payment if a journal has five orfewer loans. At that time, five uses was the breakeven point formost journals. The journal price (and processing costs and anyborrowing fees) dictates the breakeven point of use at which alibrary should subscribe (e.g., about ten uses for a US$100journal up to thirty uses for a US$500 journal and sixty uses for aUS$1,000 journals). The reason that the price increase due toincreased size of journal has not resulted in large circulationreduction is that the journals are read more because they have

July 2004 263

more articles, which means the breakeven point also increases.There are many factors that enter into collection developmentpolicies, but the cost per use of a journal or a package of journalsshould be carefully considered in the future.

FOLLOW-UP READING

King, Donald W., Sarah Aerni, Fern Brody, Matt Herbison, and PaulKohberger. Comparative Cost of the University of Pittsburgh Elec-tronic and Print Library Collections. The Sara Fine Institute for Inter-personal Behavior and Technology, 2004. http://purl.oclc.org/sfipitt/pub20040405a.King, Donald W., Sarah Aerni, Fern Brody, Matt Herbison, andAmy Knapp. The Use and Outcomes of University Library Printand Electronic Collections. The Sara Fine Institute for Interper-sonal Behavior and Technology, 2004. http://purl.oclc.org/sfipitt/pub20040405ab.Tenopir, Carol, Donald W. King, Amy Bush, ‘‘Medical Faculty’s Use ofPrint and Electronic Journals: Changes Over Time and in Comparisonwith Scientists’’, Journal of the Medical Library Association 92 (April2004) 2.

264 The Journal of Academic Librarianship

Schonfeld, Roger C., Donald W. King, Ann Okerson, Eileen GiffordFenton, ‘‘Library Periodicals Expenses: Comparison of Non-Subscrip-tion Costs of Print and Electronic Formats on a Life-Cycle Basis’’, D-Lib Magazine 10 (January 2004) 1.King, Donald W., Carol Tenopir, Carol Hansen Montgomery, Sarah E.Aerni, ‘‘Patterns of Journal Use by Faculty at Three Diverse Univer-sities’’, D-Lib Magazine 10 (October 2003) 9.Montgomery, Carol H., Donald W. King, ‘‘Comparing Library and UserRelated Costs of Print and Electronic Journal Collections: A First StepTowards a Comprehensive Analysis’’, D-Lib Magazine 10 (October2002) 8.Montgomery, Carol H., Donald W. King, ‘‘After Migration to an Elec-tronic Journal Collection: Impact on Faculty and Doctoral Students’’, D-Lib Magazine 12 (December 2002) 8.Schonfeld, Roger C., Donald W. King, Ann Okerson, Eileen GiffordFenton, ‘‘The Nonsubscription Side of Periodicals: Changes in Li-brary Operations and Costs Between Print and Electronic Formats’’,Council on Library and Information Resources, Washington D.C.,June, 2004.King, Donald W., Carol Tenopir, ‘‘An evidence-based Assessment ofthe ‘author pays’ Model’’, Nature Forum, June 25, 2004 (http://www.nature.com/nature/focus/accessdebate/26.html).