Information-Seeking Patterns

Identifying Scholarly Articles to Read

Scholarly articles are identified for reading in five basic ways: browsing, automated searches, citations found in the literature, mentions by other people, and current-awareness tools. Browsing involves personal subscriptions, library copies (including routed journals mostly in non-university organizations), and shared office or unit copies. Recent surveys show that university scientists identify about 60 percent of their readings in that way; scientists elsewhere identify 63 percent of their readings in that manner. Other methods of identification are also similar in both types of organizations: automated searches conducted online or from CD-ROM (15 percent of total readings in universities, 11 percent elsewhere); citations found in other articles and books (7 percent in each); colleagues, authors, or other persons (10 percent universities and 12 percent elsewhere); and, finally, current-awareness tools such as Current Contents and printed indexes (10 percent universities and 7 percent elsewhere). The principal difference observed between university and other organizations is that many special libraries route journals; about 16 percent of their total readings are identified in that manner. That relatively time-consuming practice is likely to be dropped when journals can be read on line.

"There has been a steady increase in library-provided articles read by scientists"

In summary, the same proportions of scientists still browse to identify scholarly articles to read, but more are using online searches. In fact, the proportion of readings identified through online searches appears to be increasing, while use of the remaining means — except for browsing — has been decreasing. Thus, to the degree possible, electronic publishing should facilitate browsing and automated searching because those are the preferred means of getting information by growing numbers of scientists, and if electronic publication is to be useful, it has to meet the readers' needs.

Table 3: How Scientists Find Articles to Read
1977 1984 1986-89 1990-93 1994-98
University Scientists Browsing 58.5% 54.1% 59.7%
Online Searching 0.7% 1.1% 14.6%
Citations 6.7% 13.1% 6.5%
Other persons 17.7% 15.3% 9.7%
Other 16.4% 16.4% 9.5%
Non-University Scientists Browsing 49.6% 67.4% 72.8% 57.1% 63.6%
Online Searching 0.2% 2.7% 2.1% 7.0% 11.1%
Citations 11.3% 6.3% 6.4% 5.4% 6.6%
Other persons 18.1% 11.3% 12.0% 16.9% 12.1%
Other 20.8% 12.4% 6.6% 13.6% 6.7%
Sources: King, McDonald, and Roderer 1981; Tenopir and King 1997; Griffiths and King 1993; and King et al. 1977-98.

Obtaining Scholarly Articles

Scientists obtain their articles primarily through personal subscriptions, nearby libraries, and other sources such as shared office collections, colleagues, and authors. Our most recent surveys indicate that university scientists rely on personal subscriptions for approximately 36 percent of readings, but personal subscriptions are used less frequently (24 percent of readings) by scientists outside universities. Scientists in universities obtain 54 percent of their readings from libraries, including reading in the university library or department library, having someone obtain photocopies of articles from the library, and ordering from interlibrary loan or a document delivery service. Approximately 11 percent of university readings come from other sources such as colleagues and authors.

Outside the university, scientists rely on libraries for more than half their readings: 63 percent of them from organization libraries, 23 percent of them from routed journals, 11 percent from interlibrary loans and document-delivery services, and 3 percent from academic, public, and government libraries. The more recent (1998) surveys of scientists working in industry and government show that nearly 10 percent of their readings are from electronic sources. Approximately the same amount of readings came from sources such as shared department or unit collections, colleagues, and authors.

Table 4: Percentage of Readings of Scholarly Scientific Journals by Source
1977 1984 1986-89 1990-93 1994-98
University Scientists Personal Subscription 60.0% 53.0% 35.5%
Library Subscription 24.8% 30.1% 53.8%
Other 15.2% 16.9% 10.7%
Non-University Scientists Personal Subscription 72.0% 66.3% 53.8% 49.0% 24.0%
Library Subscription 10.4% 20.5% 32.1% 37.3% 55.5%
Online Searching 0.1% 9.8%
Other 17.6% 13.2% 14.1% 13.6% 10.7%
Sources: King, McDonald, and Roderer 1981; Tenopir and King 1997; Griffiths and King 1993; and King et al. 1977-1998.

Thus the sources of scholarly articles read by scientists have changed dramatically over the years. There has been a steady increase in library-provided articles read by scientists in all locales. From 1977 to 1993 the proportion of those readings rose from 24.8 percent to 53.8 percent in universities; from 10.4 percent in 1977 to 55.5 percent during the 1994 to 1998 period in industry and government. That increase in library provision was largely at the expense of reading from personal subscriptions and, to a lesser degree, reading from other sources. The next table shows the decline in personal subscriptions to journals. In particular, scientists working in industry, government agencies, and national laboratories have been subscribing to fewer scholarly journals in recent years (2.44 subscriptions per scientist) than in 1977 (6.20 subscriptions).

Table 5: Average Number of Personal Subscriptions to Scholarly Journals
1977 1978-83 1984 1986-89 1990-93 1994-98
University Scientists 4.21 3.96 3.86
Non-University Scientists 6.20 4.60 4.26 3.70 2.98 2.44
Sources: King, McDonald, and Roderer 1981; Tenopir and King 1997; Griffiths and King 1993; and King et al. 1977-98.

By pooling university-survey responses with the others for 1993-98, we were able to analyze reading from journals by personal subscriptions, libraries, electronic publications, and other sources. As shown in the first column of Table 6 below, all the scientists average approximately 112 readings per year. In the second column we see that a typical scientist or engineer reads at least one article from approximately eighteen journals, up from sixteen in earlier observations (Griffiths and King 1993). Most (10.7) of those journals are provided by libraries, including interlibrary borrowing and document delivery. Personal subscriptions account for 2.9 of the journals read. The scientists read from one or two electronic journals (either on line or CD-ROM), and approximately the same number of journals provided by colleagues, authors, and other sources.

Table 6: Average Journal Readings by Article Source: 1993-98
Source of Articles Readings per Scientist Number of Journals Read per Scientist Readings per Journal per Scientist Total Readings per Journal
Personal Subscriptions 39.0 2.9 13.4 13.4
Libraries 49.7 10.7 4.6 49.2
Electronic Journals 10.5 1.5 7.0
Other 12.8 2.7 4.7
All Sources 112.0 17.8 6.3
Source: King et al. 1977-98.

Each scientist reads an average of 13.4 articles per personal-subscription journal each year (counting articles that are read more than once). Library-provided journals are read an average of 4.6 times per scientist, but collectively each journal is read an average of 49 times per year. (That figure increases to about 136 times a year when readings by students and others are included.) Each electronic journal is read about seven times per scientist per year. Other sources of journals include shared department or unit collections and copies obtained from others, such as authors and colleagues.

"In approximately 53% of the journals a scientist reads, he or she reads five or fewer articles a year"

The average readings from library-provided journals (4.6 readings) has fallen about one-half since earlier observations. We think that the decline reflects both an increase in readings identified by automated searches, which potentially broadens the range of journals read, and a reduction in library subscriptions, which has resulted in a sharp rise in the number of articles obtained from interlibrary loan and document delivery. Perhaps scientists use online searches to identify articles in journals that are otherwise scarcely read, and then they request them through interlibrary loan or document delivery. On the other hand, scientists may shy away from reading articles that their own libraries no longer subscribe to because of the cost of document delivery or the time of interlibrary loan, unless those articles are absolutely required. Such trends have appreciable implications concerning electronic journals.


We believe that there are several aspects of traditional scholarly publishing and use that must be considered in electronic publication of journals. First, there is a clear difference between universities and the rest of the world in terms of the extent to which journals are read by scientists and the purposes for which they are read (although a similarity exists in the ways articles are identified and, to a lesser degree, obtained). University libraries also differ in that they serve students and external users such as researchers from small high-tech companies. For those reasons, we have separated trends for universities and other sites.

Too, electronic publication may serve niche audiences differently than paper publication, especially for journals that have small readerships. The electronic versions may also serve libraries differently than paper versions. Similarly, journals that are infrequently and frequently read in libraries may be served in a different manner.

Another issue involves the very large, and rapidly increasing, number of articles obtained through photocopying, interlibrary loan, document delivery, preprints, and reprints.

Still another issue is that scholarly journals are read over a long period of time following publication very much like a nuclear decay curve, and older articles may not be available electronically.