Scientific Journals Are Worth Saving

Scholarly Journals Are Publishing More Research

Both the number of articles and the number of scientific scholarly journals per scientist have decreased from 1975 to 1995. At the same time, the size of journals has increased, in both the number of articles per journal and the length of individual articles, to the extent that there has been a marked increase in the number of pages published per scientist (King, McDonald, and Roderer 1981; Tenopir and King 1997).

"No matter where they are employed, scientists who read more tend to get achievement awards and other special honors"

The estimated number of journals published in the United States increased 62 percent, from 4,175 in 1975 to 6,771 in 1995, while the number of scientists more than doubled during that same period, from 2.64 million to 5.74 million in 1975 and 1995 respectively. Scientists in the United States published an estimated 312,200 articles in domestic and foreign journals in 1975 compared to 577,100 in 1995, representing a decrease of approximately 15 percent in the average number of articles published per scientist. Nevertheless, the average number of pages published per scientist increased almost 70 percent. The number of authorships per scientist has also increased because of an increase in the number of coauthors per article. For example, scientists in universities on average authored or coauthored about one article per scientist in 1977, but that number increased to 2.1 in 1995. University scientists seem to be increasing their dominance in authorship compared with authors employed elsewhere, since they authored 75 percent of all articles in 1995, up from 62 percent in 1975.

Scientists Are Reading Slightly More Articles

Recent surveys show that scientists continue to read scientific scholarly journals extensively. Scientists at the University of Tennessee read an average of 188 scholarly articles per year (1990-93); scientists in two companies in our study read an average of 106 scholarly articles annually (1994-98). While those results reflect statistical surveys from organizations that chose to participate in our survey, they match results from national surveys of scientists performed for the National Science Foundation in 1977 and 1984.

Table 1: Average Number of Scholarly Article Readings per Year
1977 1978-83 1984 1986-89 1990-93 1994-98
University Scientists 150 172 188
Non-University Scientists 90 91 99 75 80 106
Sources: King, McDonald, and Roderer 1981; Tenopir and King 1997; Griffiths and King 1993, and King et al. 1977-98.

Survey results for years other than 1977 and 1984 represent samples of scientists selected for communication studies in universities, companies, government agencies, and national laboratories as well as for studies of journal readership conducted for Science and the Journal of the National Cancer Institute. Thus readership in those years reflects the composition of scientists in organizations that volunteered to be in the study (e.g., a dominance of chemists in selected companies, medical researchers at the National Institutes of Health, and physicists in national labs) and potential journal readers (e.g., cancer researchers). Thus, the differences in reading in the two years may be an accurate predictor of what happens in other sciences, or may reflect differences in composition of the study participants rather than actual trends. However, the amount of article reading appears to be relatively constant by scientists outside of universities and may be increasing with university scientists. While university scientists tend to read more on average than other scientists, the total number of articles read must be far greater in industry and government than in universities because more scientists are employed there.

Among nine fields of science — physical sciences, mathematics and statistics, computer sciences, environmental sciences, engineering, life sciences, psychology, social sciences, and other fields — the average number of readings per article is estimated to be between 500 and 1,500. (We defined "reading" as going beyond the title and abstract to the text of the article. We count each time an article is read as a single reading.) On average, a scientific journal gets over 100,000 article readings, but the distribution is highly skewed. In other words, many scientific specialties have few scientists, and journals in such specialties therefore have relatively few readings. A small number of journals — Science is one — have large circulations with readings numbering in the millions.

Scholarly Journals Are a Critical Source of Information

In every survey we conducted, scientists were observed to read many more scholarly articles than any other type of document. Journals are read more than other types of publication because they cover a range of topics, editors attempt to maintain content quality, the information is relatively current, and the articles serve a variety of purposes. For example, scientists at the University of Tennessee, asked to indicate all the reasons they read scholarly journals, indicated that reading is done mostly for research purposes (75 percent of readings), current awareness and continuing education (58 percent), teaching (41 percent), and external communication (formal publication, 32 percent; formal presentations, 22 percent). Because 58 percent of scientists report that they read scientific journals to keep up with their fields and to continue their education, much of the reading takes place within six months of publication. Nevertheless, many of those articles are re-read later for scientific research or teaching purposes. For example, 37 percent of the scientists who read articles more than two years old were not reading them for the first time.

Those who read for research or teaching say that the information they obtain is useful: More than 30 percent say it is essential to their teaching, and nearly 40 percent say it is essential to their research. For example, one-third of the relevant readings by those university scientists are said to be essential to their teaching, and nearly 40 percent indicated that the readings are essential to their research. Scientists outside universities rated the importance of the information from scholarly journals high for every activity they performed. In fact, when compared with resources such as laboratory-instrument readouts, computer-based research, or advice from colleagues, library staff, and support staff, scientists rated the information from journals as most important or second-most important.

No matter where they are employed, scientists who read more tend to get achievement awards and other special honors. In universities, those whose teaching has been honored read about 26 percent more articles, while those honored for research read about 33 percent more articles. Outside academe, the amount of reading by achievers is even higher, 65 percent. In one company, twenty-five scientists who were considered particularly high achievers read 59 percent more articles than the other scientists and 52 percent more than non-honored cohorts from a similar specialty with equivalent degrees and years of experience. That does not necessarily suggest a cause and effect relationship between reading and achievement, but that important resource should not be denied those who hope to use it to succeed.

"Air has a low purchase value and a high use value, whereas gems have a high purchase value and a low use value"

One reason that scientists read to keep current is that the amount of recorded scientific knowledge is doubling approximately every fifteen to twenty years (Price 1963). Thus in 1998, all the knowledge recorded throughout history to the early 1980s has now doubled and will double again in the next fifteen to twenty years. The impact of that exponentially increasing amount of knowledge is that when scientists are graduated from college they have been exposed to only a fraction of the knowledge they will need in their careers. In fact, that knowledge base is only approximately one-sixth of the knowledge that they must later master. Thus, they must either continue to read or risk falling behind in their research and teaching.

Detailed Value Analysis of Scholarly Journals

We examined two aspects of the value of information in scholarly journals: the purchase value and the use value. The purchase value is what users are willing to pay for the information; the use value reflects benefits gained from the use of the information. The two values can vary considerably from product to product. For example, air has a low purchase value and a high use value, whereas gems have a high purchase value and a low use value. The purchase value of information provided by scientific scholarly journals has two components: the money exchanged for subscriptions and the time scientists expend in obtaining and reading the information. The use value can be measured in terms of how the information affects the performance of scientists (Machlup 1979).

In our most recent surveys we found that the average university scientist purchased 3.9 journal subscriptions a year, and the average scientist outside the university purchased 2.4 subscriptions a year. The average university scientist spent 182 hours a year reading articles; the average scientist outside the university spent 64 hours reading articles (and another 12 hours just obtaining those articles). Even if the average journal subscription is $100 a year, and the average scientist makes $50 an hour, the value of the time spent getting and reading journals is greater than the cost of the journals. Since a scientist's time is a scarce and valuable resource, his or her willingness to expend it on information found in articles in scholarly publications is an indicator of the value of the information to the scientist.

The average amount of time spent reading scholarly articles has remained relatively constant over the years as shown below. The times are estimated by multiplying the average time per reading, by the estimated annual number of readings. The time spent reading is validated from other methods of observing reading time (King, McDonald, and Roderer 1981; Griffiths and King 1993). We do not know why the 1984 survey estimates of average time per reading is higher than other years, since the questions were worded the same in all of the surveys.

Table 2: Average Time Spent Reading Scholarly Scientific Journals
1977 1978-83 1984 1986-89 1990-93 1994-98
University Scientists 120 hours 259 hours 182 hours
Non-University Scientists 70 hours 86 hours 101 hours 70 hours 68 hours 64 hours
Sources: King, McDonald, and Roderer 1981; Tenopir and King 1997; Griffiths and King 1993; and King et al. 1977-98.

The use value is examined through several indicators, including perceived effects on scientists' activities and their productivity. Nearly all university readers (95 percent) indicated some favorable outcomes from some of their readings. They said two-thirds of the readings improved the quality of teaching, research, or other activity for which they read the article. They also indicated that reading helped them perform the activity better (33 percent of readings), faster (14 percent of readings), or saved them time or money (16 percent of readings).

Scientists in industry and government said approximately two-thirds of their readings improved the quality of the activity for which the article was read, with ratings of quality (ranked from 1 to 7) increasing from an average of 4.04 to 5.82 following the readings (Griffiths and King 1993). The readings also let them perform activities faster (32 percent of readings), reinforce hypotheses or confidence in work (42 percent), initiate ideas for research (26 percent), and broaden (23 percent) or narrow options (6 percent) concerning research. We derived five indicators of productivity: number of formal records of research (e.g., lab notes), number of consultations, number of presentations, number of written proposals or plans, and number of formal publications written. Each output measure was divided by the amount of time spent (e.g., doing research). We found that each of the five indicators correlated positively with the amount of reading; in fact, they were statistically significant in every case.

Another indicator of use value is the scientists' perception of savings achieved as a result of reading articles. Scientists told us that about 26 percent of the readings resulted in saving time or other resources, with the average saving estimated at approximately $300 per reading (including those cases in which there are no savings). (The average does not imply a typical saving, since some readings have no savings and only a few readings — 1 percent or 2 percent — account for nearly all the savings.) Savings are said to result from avoiding having to do some primary research; stopping an unproductive line of research; or modifying research, analysis, or engineering design.