Donald W. King, Research Professor
Carol Tenopir, Professor
Carol Hansen Montgomery, Dean of Libraries
Sarah E. Aerni, Special Projects Librarian
University libraries are rapidly moving toward electronic journal collections. Readership surveys at three universities with different levels of electronic journal implementation demonstrate how transition to electronic journal collections affects use patterns of faculty and staff. The University of Tennessee was in a transitional phase when the survey was done (2000), the University of Pittsburgh had acquired a large electronic journal collection, but with some duplication with print journals (2003), and Drexel University had migrated to nearly all electronic journals (2002). Although faculty use of personal print subscriptions remains significant, electronic personal subscriptions are used only infrequently by faculty even though this is an option available to them. On the other hand, electronic journal use is very high when available in library collections. Twenty-five year trends of reading by university scientists show substantial increases in average amount of reading with nearly all of this increase coming from library collections. The likely increase in reading from library collections is due in part to a decline in personal subscriptions and increased online bibliographic searching coupled with increased availability of the library collections and, recently, enlarged electronic journal collections. Scientists appear to be more advanced in their use of electronic journals than other faculty, but changes are taking place within all faculty disciplines.
A series of readership surveys conducted by the authors over the past 25 years provide a useful examination of information-seeking and reading patterns in U.S. universities. These surveys began in 1977 with a national survey of scientists performed under a National Science Foundation (NSF) contract [King et al, 1981]; it was replicated in 1984, again under NSF sponsorship. While surveys were conducted every year since 1984 in non-university settings, it was not until 1993 that a readership study focused on a university setting [Belefant-Miller and King, 2001]. All of the readership surveys from 1977 to 1998 were summarized by Tenopir and King in an attempt to describe changing use patterns leading up to the evolution of e-journal publishing [Tenopir and King, 2000]. In 2000, the surveys began to focus on changing patterns of journal use due to e-journal publishing and their advanced features [Tenopir et al., 2003]. In addition to surveys of American Astronomical Society members (2001) and Oak Ridge National Laboratory professionals (2000), three studies partially discussed in this article, involved universities: University of Tennessee in 2000 [Tenopir and King, 2002], Drexel University in 2002 [King and Montgomery, 2002] and University of Pittsburgh in 2003.
This article presents results of the three recent university surveys to show extent of journal reading, source of articles read, the extent to which print and electronic formats are used and how readers learned about the articles. Because 25-year reading trends are available, at least for university scientists, the relevant university survey responses are extracted to show how reading has increased over this time period and the extent to which e-journals, library access and online bibliographic searching have contributed to increased reading by university scientists. These factors may portend reading trends in other fields in universities in the future.
Circumstances in the three universities are quite different. The University of Tennessee is a large university (26,000 students) that has scientific ties with Oak Ridge National Laboratories. In 2000, when the survey was conducted, their Hodges Library was in a transitional phase in terms of journal formats available to users. Drexel University's Hagerty Library, on the other hand, had migrated to a nearly all-electronic acquisition by 2002 when the survey was performed. Drexel is a smaller university and had about 10,000 students and 500 faculty at the time of the survey (see [Montgomery and King, 2002], for a description of this transition to an electronic journal collection). Both Tennessee and Drexel rely heavily on a central library system. The University of Pittsburgh is much different in that its University Library System (ULS) serves five campuses and has 19 large departmental libraries in addition to the central Hillman Library. The 2003 Pittsburgh survey included all faculty except law and health sciences serving 32,660 students at these sites. Pittsburgh has recently acquired a large collection of e-journals, many of which duplicate print acquisition. Thus, use patterns observed in these three institutions reflect their differences, although all three have strong scientific educational and research facilities.
Previous Studies of Journal Use in Universities
There are three recent reviews of the literature on journal use. One is a review of past journal use [King and Tenopir, 2001], another emphasizes changing patterns [Kling and Callahan, 2003] and the third summarizes cross-university and individual university studies [Tenopir, 2003]. Some multi-university studies focus on how faculty use libraries, online resources and electronic journals [Healy et al., 2002; Friedlander 2002; Institute for the Future, 2002; Lawal, 2002; and Tenopir and King, 2002]. Other recent studies show that faculty in the sciences tend to prefer and read more in electronic journals or e-prints than do humanists or those in the social sciences [Brown, 1999; Brown, 2003; Dillon and Hahn, 2002; Tenner and Yang, 1999; and Cochenour and Moothart, 2003]. Most faculty want non-core journal titles in an electronic format, with core titles in both print and electronic format [Dillon and Hahn, 2002]. Convenience, with access from the desktop is especially important to faculty [Maughan, 1999]. These studies tend to confirm results presented in this article.
This article discusses findings from two sets of readership surveys involving universities. The first set includes six surveys of scientists located in universities in the U.S. Two surveys (1977 and 1984) involved national samples of scientists. These two surveys yielded responses of 2,350 and 865 respectively, of which about one half were located in universities. The response rates for these surveys were over 60%. Readership surveys were also performed in the University of Tennessee (1993 and 2000), Drexel University (2002) and the University of Pittsburgh (2003). Only the 389 responses from science faculty were used in this article. These surveys describe 25-year trends of readership patterns of scientists.
The second set of surveys involves all faculty at the three universities (University of Tennessee- 1993, 2000, Drexel University- 2002 and University of Pittsburgh- 2003). The total number of responses for faculty and staff was 527, and the surveys in these universities yielded between 18 and 50% response rates. Details of the survey methods are given in [Griffiths and King, 1991].
Journal Use by University Faculty and Staff
Journal use is measured by the following question: "In the past month (30 days), approximately how many articles have you read? Articles can include those found in journal issues, Web sites or separate copies such as pre-prints, reprints and other electronic or paper copies. Reading is defined as going beyond the table of contents, title and abstract to the body of the article." Such reading often involves re-reading of an article. In fact, about 22% of the reading reported below involves re-reading of an article. The estimates of average reading per faculty or staff member  below reflect projection to an entire year.
The average annual amount of reading per person is displayed in Figure 1 below.
While there is some difference in average amount of reading among the three universities, scientists (including engineers and social scientists) tend to read more than non-scientists  in all three sites (Tennessee 201 vs. 186 readings, Pittsburgh 226 vs. 185, Drexel 214 vs. 140). Nevertheless, reading by faculty is substantial and, perhaps, increasing as shown in the section on 25-year trends in university scientists' use patterns.
Source of the Articles Read
Articles are read from many sources including from personal subscriptions, library collections, separate copies of preprints and reprints, and copies provided by colleagues and author websites, among other sources. They are now read in several formats including print and electronic. In Table 1 below, we show the data from the three universities that indicate the average amount and proportion of reading obtained from these sources and the format used.
Sample Size: Tennessee (n=96), Pittsburgh (n=209), Drexel (n=92)
The transitional phase at the University of Tennessee is evident in that the amount and proportion of reading from library collections and electronic formats are well below that observed at Pittsburgh and Drexel. Availability of library electronic collections may contribute to increased use of university libraries, and more is said about this later. It is noted that faculty reported reading from "free Web journals" at both Tennessee and Pittsburgh, but not Drexel. Drexel devoted substantial effort in publicizing the migration to an electronic collection over four years and in training users to use this collection. It may be that some faculty at the other two universities are unaware that some of the "free" journals are in fact provided through the library collections. Thus, even though there are free journals on the Web, there also may be a "branding" problem for some university libraries in providing access to their electronic journal collections.
One of the most interesting observations is that reading from personal subscriptions is almost exclusively from print, even though most journals are also available in electronic format. On the other hand, most reading from the library collection is in electronic format at the two universities where electronic collections are well established. One reason for the latter phenomenon is that faculty can access their library articles in their offices or at home. We estimate that this saves them about 15 minutes per reading or 20 hours per year per faculty member. Most reading from personal subscriptions is from recently published articles (85% of readings were published in the past two years), which largely involves browsing to keep up with the literature. The average amount of reading from personal subscriptions among the universities ranges from about 70 to 90 readings per faculty member or about 20 to 25 readings per personal subscription. Most of these subscriptions are part of society membership, but some are purchased from grant funds.
Science faculty are observed to rely much more on library collections and separate copies of articles than non-science faculty. They also read twice as many library-provided articles in electronic formats, again suggesting that availability of electronic journals in library collections may lead to more library use.
Format of Articles Read by Faculty
The format of articles read is shown in Table 2 below.
Sample Size Total = 397, Scientists = 300, Non-Scientists =97, Tennessee = 96, Pittsburgh = 209,
These results include the format of separate copies of articles. About 40% of reading is now from electronic format. Faculty from Tennessee were less likely to read e-journals during their library's transitional phase (2000) and scientists clearly lead non-scientists in the use of e-journals. However, it may be that non-scientists will eventually follow suit.
How Faculty Learn about the Articles They Read
There are many ways that faculty identify and locate the articles they read. The most frequent (about one-half of reading) is by browsing recently published issues of journals (personal, library or other source). This browsing is usually done for current awareness and keeping up with the literature or while conducting background research. Increasingly, faculty may identify relevant articles (about a quarter of reading) by online searching. Such searches include traditional abstracting and indexing (A & I) databases, Web search engines, searchable online journal collections or electronic current awareness services. About 70% of these searches are from A & I databases, particularly by scientists. However, once identified, these articles must be located and obtained before they can be read. Because libraries are known to provide access to a broad range of journals, most of these articles are obtained from library collections and are obtained in electronic format (sometimes through a direct link). Other means of learning about articles is from citations found in other publications (11% of readings), from another person, e.g., a colleague (13%), or some other means (3%). These ways of identifying articles also require gaining access to the articles (although when colleagues inform about an article, they often also provide the article in photocopy or e-format).
Table 3 shows the methods initially used by faculty to learn about articles they read.
Sample Size Total = 397, Tennessee = 96, Pittsburgh = 209, Drexel = 92
The articles' identification patterns do not vary much among faculty from the three universities.
Age of Articles Read by Faculty
Most reading takes place soon after articles are published. However, older articles are often read as well. For example, in the university surveys, the oldest article read was published over 50 years ago (1949). Some articles may have been initially identified through browsing a recently published issue, but read again from a library copy many years later. Many older articles are now available in electronic format through JSTOR, society journal backfiles (e.g., American Astronomical Society, American Chemical Society) and some commercial publishers (e.g., Elsevier via Science Direct). However, our surveys have yet to demonstrate that this has had an appreciable impact on reading older articles, except to save readers time in obtaining them from library electronic collections.
Table 4 below shows the distribution of the age of articles read by faculty at the three universities.
Sample Size: Total = 397, Tennessee = 96, Pittsburgh = 209, Drexel = 92
The distribution is very similar for Pittsburgh and Drexel, but the age of articles read at Tennessee appears to be newer. This could be due to Tennessee having fewer older articles available in electronic format, but this assertion cannot be validated.
Table 5 displays the source of articles read by the age of the articles.
Sample Size: Total = 397, Scientists = 300, Non-Scientists =97
Recently published articles tend to be obtained from personal subscriptions, but older articles are obtained more frequently from library collections.
25-Year Trends in University Scientists' Use Patterns
Since 1977 there have been 41 readership surveys of professionals in various organizations of which about 2,000 of the 15,000 scientists' responses were from university scientists. Six of these surveys had sufficient responses from university scientists to show 25-year trends in their use patterns. Figure 2 below shows average annual reading per university scientist in 1977, 1984, 1993 and collectively 2000, 2002 and 2003.
These results suggest that university scientists have tended to read more in recent years than in 1977 . The same is true with non-university scientists as well [Tenopir and King, 2000]. The surveys do not reveal why the amount of reading appears to be increasing. The data show that it may be a phenomenon of taking much less time per reading. This would be an interesting question to pursue in future studies.
We next examine the source of the articles read. Figure 3 shows the trend in amount of reading from library collections and from all other sources.
The net growth of reading is 66 readings from 1977 to recent years and it is 64 readings from library collections over the same time period . This suggests that library collections have absorbed this large growth in reading. A parallel observation is that university scientists read from a much broader range of journals. In 1977 they read at least one article from 13 journals, but now read from over twice that number of journals .
There are several plausible reasons why libraries are being used more. First, university scientists subscribe to fewer journals (i.e., 4.2 subscriptions in 1977 to about 3.5 currently) and the estimated difference of about 15 readings from these sources partially helps explain the increase in use of material from library collections. Another reason is that online searching of bibliographic databases has increased substantially. Only about three articles per scientist were identified in this manner in 1977 compared with nearly 50 articles currently. While some online bibliographic searches involve current browsing, many articles identified are older articles that are obtained from library collections. The breadth of reading and corresponding increased use of library collections are also undoubtedly due to access to enlarged electronic library collections. For example, the set of Drexel's print library collections was about 2,000 journals and the electronic library collection is nearly 10,000 unique titles  and preliminary analysis shows significant use of electronic titles not subscribed to previously in print.
The current surveys provide an estimate of 216 readings per science faculty. The average amount of time spent per reading is 35.4 minutes per reading or about 127 annual hours per scientist. These two results are indicators of the value scientists place on the information read since they would not choose to spend this valuable time if the information was not worthwhile. It is noted that reading from library collections involves a longer time (39.4 minutes per reading) versus other sources (32.7 minutes per reading), thus indicating that the reading from library collections may have greater value.
The principal purpose for which information is read is most frequently said to be for primary research (35% of readings) with teaching being the second most frequently mentioned (19%). The scientists were asked to rate the importance of the information content to achieving the principal purpose. The average importance rating (from 1- not at all important to 7- absolutely essential) was 5.1. However, articles read from library collections were rated a 5.5 on this scale versus 4.7 for all other sources, providing additional evidence of the usefulness of library collections. The scientists were also asked to indicate the ways reading affected the purpose for reading (basically attempting to gauge the impact of the reading). The most frequently mentioned was that reading inspired new thinking or ideas (39% of readings); narrowed, broadened or changed the focus of research, teaching, etc (31%); led the reader to new authors or data sources (22%); and confirmed suspicions (19%). These proportions tend to be higher for articles provided by the library. For example, 45% of all readings from library-provided articles inspired new thinking versus about 31% from other sources.
Another indicator of the usefulness of reading journals is that science faculty whose work has been recognized through awards or special recognition in the past two years tend to read more than others. For example, those whose work has been recognized averaged 288 readings per year while the others averaged 203 readings. Faculty who received awards mentioned reading much more from library collections than from other sources.
Finally, we asked faculty whose last reading was from a library-provided article whether they would have obtained the article if there had not been access from the library and, if so, where would they have obtained it and how much more would it have cost them in their time or money. They said they would not bother obtaining the article for about 18% of the readings. In cases where they "would have" obtained articles elsewhere, we compared the time that it took them to obtain the article from the library with the time and cost "that would have been" required to obtain the information elsewhere. This comparison showed that it would take the reader an average of about 15 minutes and $8 more per reading. Thus, the library saves each scientist about 21 hours a year plus $660 per year in additional costs.
On the other hand, we observed very little difference in the indicators of usefulness and value in reading from electronic and print formats. This may be due to the fact that information content is essentially the same in the two formats and it is the content that yields value.
Conclusions from these surveys are as follows:
While science faculty have led the way to electronic journal use, others seem likely to follow as electronic journals become more widely available to them.
 Faculty refers to both faculty and staff through the remainder of the article.
 The reading by faculty excludes medicine and law. Medical faculty reading is found in [Tenopir et al., (2003) and detailed discussion of reading by engineering faculty is found in [Tenopir and King, 2003].
 Survey questions regarding reading amounts have not changed in all 41 readership surveys except to reflect electronic format options.
 A similar trend is observed for non-university scientists.
 The distribution of reading per journal is highly skewed. That is, a few journals are well read (often personal subscriptions or expensive journals found in library collections) and many are infrequently read.
Brown, Cecelia M. 1999. Information Seeking Behavior of Scientists in the Electronic Information Age: Astronomers, Chemists, Mathematicians, and Physicists. Journal of the American Society for Information Science, 50 (1999): 929-43.
Brown, Cecelia M. 2003. The Role of Electronic Preprints in Chemical Communication: Analysis of Citation, Usage and Acceptance In the Journal Literature. Journal of the American Society for Information Science and Technology, 54 (5): 362-371.
Dillon, Irma F. and Karla L. Hahn. 2002. Are Researchers Ready for the Electronic- Only Journal Collection? Results of a Survey at the University of Maryland. Portal: Libraries and the Academy, 2 (3): 375-390.
Friedlander, Amy. (2002). Dimensions and Use of the Scholarly Information Environment: Introduction to a Data Set. Washington D.C.: Council on Library and Information Resources. Available at: <http://www.clir.org/pubs/reports/pub110/contents.html>
King, D.W. and C. Tenopir. 2001. "Using and Reading Scholarly Literature." In: Annual Review of Information Science and Technology Edited by Martha Williams. Vol. 34. Medford, NJ: Information Today, Inc..
King, D. W., D. D. McDonald and N. K. Roderer. 1981. Scientific Journals in the United States: Their Production, Use and Economics. Stroudsburg, PA: Hutchinson Ross Publishing Company (Division of Academic Press).
King, D. W. and Montgomery, C.H. 2002. After migration to an electronic journal collection: Impact on faculty and doctoral students. D-Lib Magazine, 8:212. Available at: <doi:10.1045/december2002-king>.
Kling, R. and Callahan, E. 2003. Electronic Journals, the Internet and Scholarly Communication. In Cronin, B., Ed., Annual Review of Information Science and Technology. 37. Medford, NJ. Information Today, Inc.
Lawal, Ibironke. 2002. Scholarly Communication: The Use and Non-Use of E-Print Archives for the Dissemination of Scientific Information. Issues in Science and Technology Librarianship, 26. Available at: <http://www.istl.org/02-fall/oarticle3.html>.
Maughan, P. D. (1999). Library resources and services: a cross-disciplinary survey of faculty and graduate student use and satisfaction. Journal of Academic Librarianship, 25(5), September 1999, 354-366.
Montgomery, C. H. and King, D.W. 2002. Comparing library and user related costs of print and electronic journal collections: A first step towards a comprehensive analysis. D-Lib Magazine, 8:10. Available at <doi:10.1045/october2002-montgomery>
Tenopir, C., King, D. W. , Boyce, P., Grayson, M., Zhang, Y., and Ebuen, M. (2003). Patterns of journal use by scientists through three evolutionary phases. D-Lib Magazine, 9:5. Available at <doi:10.1045/may2003-king>.
Copyright © Donald W. King, Carol Tenopir, Carol Hansen Montgomery, and Sarah E. Aerni