This article is adapted from summary information in Kevin Guthrie's paper "Revitalizing Older Published Literature: Preliminary Lessons from the Use of JSTOR", presented at the The Economics and Usage of Digital Library Collections Conference held at the University of Michigan in March 2000

As part of our not-for-profit mission, one of our objectives is to contribute to the understanding of how scholarly materials in electronic formats are being used. That was a key reason we developed the web statistics reporting system for libraries and publishers. Now that JSTOR has been available at some licensed institutions for as long as three years, we have begun to accumulate enough data to initiate questions about the impact of JSTOR on the use of older literature. For example: Do scholars and students utilize older articles? Does the usefulness of older literature vary by academic discipline? Are the digitized versions of older articles used more frequently than their paper counterparts? And do these data provide information that is useful for selecting scholarly materials to be digitized?

We are only at the initial stages of analysis, and for many of these questions we must collect much more data for any assertions to have statistical validity. Still, what we are finding already opens a fascinating window into some surprising usage trends, and points to hypotheses in five key areas.

1. The availability of older journal articles in electronic form through JSTOR seems to have increased the use of the older articles at participating sites.

In 1996, prior to widespread availability of JSTOR, we conducted a usage survey of ten JSTOR journals (in their paper format) at 6 colleges and universities. The mechanisms for counting uses of these paper journals were far from perfect, but they did give us a very rough sense of the extent of usage of these materials. Working cooperatively with librarians at these institutions, we recorded a total of 692 uses of the ten paper journals over a three month period in 1996. We then counted the number of uses in JSTOR of these same titles at the same six institutions during the last three months of 1999. A total of 7,696 articles were viewed and 4,885 were printed over the course of three months.

Another way to address the question of whether JSTOR is increasing use of these older materials is to evaluate the growth in usage. Judging from conversations with librarians, it is a safe assumption that the use of older journal articles (in paper form) was not growing prior to their being digitized. By contrast, looking at the usage of JSTOR at the 82 sites that have had access to the resource since early 1997, one discovers that aggregate accesses at these institutions increased by a factor of 3.4 times from 1997-1998 and 2.5 times from 1998-1999. The cumulative growth in usage of the JSTOR database over this two-year period was an astonishing 740%.

2. Researchers and students value the interdisciplinary nature of JSTOR.

Another notable finding was that researchers are taking advantage of JSTOR's cross-title and interdisciplinary capabilities. For example, after sampling 68,000 searches in a single week of JSTOR use, we learned that approximately 90% of the searches specified more than a single title. In addition, JSTOR's ability to search across disciplinary clusters seems important to users. Out of 58,000 recent cluster-specific searches, 69% specified more than one cluster. As JSTOR adds new content in existing fields, and begins digitizing journals in additional academic disciplines, the interdisciplinary nature of JSTOR is likely to become even more important to users.

3. Older literature remains valuable in many fields.

One of the goals of the study was to take an initial snapshot of the relative value of older literature in the academic fields included in JSTOR. As a first estimation of this value, we looked at the top ten most frequently used articles (in terms of the number of times that the article has been viewed and/or printed) and noted the age of these articles.

In most of the major fields included in JSTOR, the articles in the top ten were older than one might have expected they would be. In economics, for example, the average age of the top ten articles most frequently printed and viewed was 13 years. More dramatically, in the field of mathematics, the average age of the most used articles was 32 years. These data are by no means conclusive, as some of the JSTOR journals have only been digitized relatively recently, but the early findings seem to contradict existing assumptions about the value of older literature.

4. Citation data alone do not reliably predict electronic usage.

Judging by the most-used articles in JSTOR, citations and usage do not correlate closely, suggesting that citations should not be used as the sole factor in selecting content to be digitized. To give just one example, the most frequently viewed article from one of the top journals in the economics collection has rarely been cited in other articles. The article, published in 1973, was cited only fourteen times between 1974 and 1999. Nevertheless, this article has been viewed 1,895 times and printed 1,402 times since it was made available in JSTOR, making it the 4th most-used article in economics. (note: Economics is the most-used collection in the JSTOR database, accounting for approximately 18% of total accesses). One interesting question raised by these data is whether the availability of these older articles in electronic form will increase their citation frequency and lengthen their citation "half-life." It is far too early to begin analyzing this question, but it is worth following.

5. The concept of "value" for research articles needs to be clearly understood as libraries consider acquisition and cancellation decisions for electronic content.

Increasingly, one hears that usage data should be used more aggressively by librarians in acquisition and cancellation decisions for current journal subscriptions. This makes perfect sense, as it goes to one aspect of the value of the journal to the constituency of that library. But it is important to recognize that usage is only one aspect of the value, not the entire value. Citations and citation impact factors reflect another kind of value.

The fact that top used articles in JSTOR may be infrequently cited, or that top-cited articles may be infrequently used, does not prove that one or the other is more important; rather, it indicates that both components must be considered. An article that gets assigned to an Economics 101 class at a large university will generate large numbers in the JSTOR statistics, but that high usage does not necessarily reflect the importance of the article to research and the future intellectual development of the field. The same could be said for the value of usage statistics in faculty tenure evaluations.

Usage statistics provide important information about the value of a journal on a campus, but they are more likely to reflect the value of the journal as a teaching resource than as a research resource. Both perspectives should be taken into account when using these data to help make journal subscription decisions.

A complete copy of this paper is available at:
http://www.jstor.org/about/preliminarylessons.html.

Last updated on September 8, 2006