American Science Plateau

The National Science Foundation released a pair of reports on Wednesday about the decline in the American share of published articles in science and engineering worldwide -- not exactly surprising in light of the growing influence of scientists from Asia and Europe.

What the studies found, however, was that besides the well-known decrease in the relative share of journal articles originating from the United States, there was a slowdown in absolute numbers as well. This "plateau," as the reports call it, began in the early 1990s and stands in marked contrast to at least the two previous decades' worth of American research.

The flattening of growth in science and engineering publishing -- it has "essentially remained constant since 1992," according to the first report -- remains partly a mystery. The report also asserts that there hasn't been a corresponding decrease in "resource inputs," such as funding and research staff, that might stunt the growth of American output in scholarly journals. And the plateau is seen across multiple disciplines within the sciences.

Still, the reports, part of a projected three-part study by the NSF's Division of Science Resources Statistics, highlight some issues that could be behind the trend, such as increasing collaboration and institutional differences between countries. The first, called "Changing U.S. Output of Scientific Articles: 1988–2003", takes a look at the hard numbers. The second, "The Changing Research and Publication Environment in American Research Universities", is based on extended interviews with researchers and administrators at nine major research universities and offers some possible explanations for the plateau in publishing volume.

(The universities surveyed were: Boston University, the California Institute of Technology, Duke University, Harvard University, Stanford University, the University of California at Los Angeles, the University of Chicago, the University of Illinois at Urbana-Champaign and the University of North Carolina at Chapel Hill.)

The first report tracked scientific journals, notes and reviews from the database Thomson ISI, now known as Thomson Scientific. One methodological problem encountered by the authors for their comparisons was that the journals in the database change from year to year, as new publications are added and a smaller number of others fall off the radar of the scientific community. Rather than stick with a fixed subset of journals that would stay the same each year, the researchers opted to use the "expanding set" in its comparisons, since the other method could potentially ignore new journals created for emerging subfields. Either way, the report said, the basic findings remain the same.

The flat output of U.S.-originated articles in the sciences and engineering, as well as continued growth in other nations in Europe and Asia, have led to the well-publicized trend of a declining U.S. share of scientific research papers published worldwide -- from 37 percent in 1992 to 30 percent in 2003. This raises the question of whether a decline in federal funding has affected American output, but the first report shows otherwise.

"Trends in federal funding by field did not generally coincide with trends in article output in broadly comparable fields," it notes. The table below compares federal funding and the growth in number of articles during two different periods of time, with a two-year delay to allow for funding "inputs" to be processed into published "outputs."

Average Annual Growth in Research Funding and Article Output

If funding isn't the answer -- or at least, not the only answer -- then an explanation could lie in an increasing trend toward collaboration between researchers, as interviews in the second report suggest. The decline in three fields in particular -- clinical medicine, biomedical research and the earth/space sciences -- are singled out as probably the result of this trend, since "U.S. authors were increasing their collaboration with the rest of the world."

"If you have more multidisciplinary research being done, and it’s not being done by individual investigators but more by teams … you likely end up with more people on a paper and less papers," said Tobin Smith, the associate director of federal relations for the Association of American Universities.

Still, Smith said that funding could still be a factor. The physical sciences, for example, have been receiving flat federal funding that, with inflation, is "actually decreasing," he said. According to the reports, "the U.S. share of the world’s [science & engineering] articles remained relatively more robust in biomedical fields than in the physical sciences and engineering, where share declines tended to be greatest and output statistics tended to lag. In fields where U.S. shares of world article output dropped least, the United States was increasing its rate of international collaboration relatively quickly and thus was increasingly sharing credit with other countries."

The second report (from its various interviews), along with Smith of the AAU, also suggested several other reasons for the American publishing plateau:

• Pressures at American universities may be geared more toward increasing quality rather than quantity. In a similar fashion, the metrics used by the NSF's researchers -- "fractional counts" versus "whole counts" for papers with multiple authors, which can alter how much weight is assigned to each country involved in a collaborative journal article -- might not sufficiently capture the priorities of U.S. research institutions.
• The process of writing and applying for grants could be affecting American productivity as compared with that of other nations: "Those interviewed repeatedly told the study team that researchers were putting much more effort into securing funding than they had in the past."
• The equipment and technology needed for scientific research is more expensive and requires more resources than such work did even 20 years ago, Smith suggested, meaning that even steady increases of funding or personnel might not be enough.
• Regulations could also be a factor: "Many researchers reported that they spent significantly more time complying with government regulations than they had in the past."

A third report is planned that will explore the various factors in detail by analyzing the various "inputs" and "outputs" to determine which might be most relevant to the overall trend.