An Unexplained Pause

THE NUMBER OF SCIENCE and engineering articles published by U.S.-based authors in the world's major peer-reviewed journals rose from 177,700 in 1988 to 211,200 in 2003. This was a modest increase of 33,500, or 19%, over the 15 years.

The number of articles from authors based elsewhere grew from 288,700 to 487,500 over the same period for a more substantial gain of 198,800, or 69%. Overall, the U.S. accounted for just 14% of the total increase in articles over these years and its share of the world total fell from 38% to 30%.

These figures have caught the eye of the National Science Foundation. The agency is not particularly alarmed or surprised that growth was faster abroad over these years, as the U.S. remains the world's dominant single science power by a large margin.

In light of the steady rise in R&D funding, research personnel, and other research inputs in this country, however, NSF apparently is concerned about the unprecedented and almost total lack of growth in publication of U.S. articles that began in the early '90s. There were 194,000 U.S.-based science and engineering articles published in 1991 and 195,800 in 2002.

The trends are similar for chemistry articles. U.S.-based researchers accounted for only a little less than 10% of the 1988-to-2003 increase in papers worldwide. The U.S. share of chemistry articles overall slipped from 23% to 19%.

These data are based on counts of scientific and engineering articles, notes, and reviews published in the more than 5,000 journals tracked by Thomson Scientific and indexed in Science Citation Index and Social Sciences Citation Index.

It may be that these divergences in U.S. and non-U.S. growth rates have peaked and are now on the decline. The number of articles abstracted by Chemical Abstracts Service grew by 12.5% worldwide between 2003 and 2005, with U.S. papers increasing by 11.9% and non-U.S. papers by 12.6%.

Similarly, the U.S./non-U.S. gap in the rates of growth in the number of articles published in American Chemical Society journals has been closing. Between 1999 and 2001, the number of non-U.S. articles grew by 10.7%, while that of U.S. papers grew by a much slower 2.7%. Between 2003 and 2005, these gains were larger and more even: 24.0% for non-U.S. articles and 16.2% for U.S. articles.

Even if a revival of U.S. journal article publication is afoot, the 1991-to-2002 plateau raises questions that NSF's Division of Science Resources Statistics is in the throes of a three-report study to answer.

The first two reports were published recently. One is a statistical survey: "Changing U.S. Output of Scientific Articles: 1988-2003." The other, "The Changing Research and Publication Environment in American Research Universities," is based on qualitative assessments of what has been happening in academic research over the past 15 years or so by experienced researchers and research administrators at nine top-tier research universities. The final report will be a more quantitative analysis.

The first report indicates that the plateau in U.S. publications was pervasive. The numbers of chemistry, biology, and physics articles published between 1991 and 2002 decreased 4.3%, 5.6%, and 15.6%, respectively. The clinical medicine/biomedical research category had a nominal publishing gain of less than 5% over those 11 years, mathematics articles were up by 5.1%, engineering/technology articles by 6.5%, and earth science articles by 16.6%.

The second report, which is in the form of a working paper, is based on the output from focus groups and individual interviews at the nine universities involved. Among the host of comments, insights, and impressions offered were the following:

• ◾ Peer-reviewed articles are the major vehicle by which research findings are validated.
• ◾ The centrality of journal articles for assigning credit is unchallenged.
• ◾ Pressure on scientific productivity in the U.S. is increasing with an emphasis on enhanced quality. Overseas, the emphasis remains largely on quantity.
• ◾ There is increasing interdisciplinary and interinstitutional research collaboration that can result in more people, money, and time being needed to produce a publishable article.
• ◾ Competitive institutional demands are not diverting faculty from research, but the increasing time and effort needed to acquire funding may lower U.S. research productivity.
• ◾ Changes in the role of peer-reviewed articles are unlikely to account for the trends observed in article counts.

These factors are presumably among those that the final NSF report will explore further and attempt to quantify. Other questions addressed will likely include: What has been the impact of the increasing median age of the research workforce? Has the increasing number of women researchers had any impact? Have the minimal growth for published articles and the increasing resources being applied to research combined to boost the scientific content of articles?

Whatever it finds, the final report should make intriguing reading.