Contents

---

Executive Summary

Purpose and Scope

Findings

Conclusion

Appendix A. Universities Visited

Appendix B. Method

Appendix C. Interview Guides

Suggested Citation, Acknowledgments

Conclusion

As explained at the outset of this report, SRS intended its university visits to be exploratory. SRS does not believe that data such as these can yield either definitive descriptions of how academic S&E research has been changing or definitive explanations of the causes of the publication trends observed in the quantitative data. Moreover, data from interviews and meetings with knowledgeable informants are not very useful, even in an exploratory way, for considering some possible explanations (e.g., that article counts are affected by an aging researcher workforce that produces fewer articles per researcher).

Nonetheless, it may be useful, in concluding, to bring the descriptive generalizations from the university visits to bear on the trends SRS ultimately hopes can be explained and to identify issues for which more extensive or systematic data collection or analysis might be warranted.

The findings from the university visits suggest that changes in the role of peer-reviewed journals are unlikely to account for the trends observed in article counts. Throughout the period studied, peer-reviewed journals remained the major vehicle by which research findings were validated and scientists obtained credit for their contributions. According to the study informants, data on article counts are unlikely to mask or distort real changes in scholarly output, except, possibly, in computer sciences. If U.S. researchers figure less prominently in the journal literature, the reason does not appear to be because they are reporting their findings in ways that bypass the journals.

Because this study was prompted by the difference between publication trends in the United States and other countries, observations that may relate to national differences in research output are especially salient. Conversely, changes in research and publication practices that seem more likely to have relatively uniform effects on output worldwide, such as changes in computer software that make manuscript preparation easier, hold little promise for helping to explain variation among nations.

Perhaps the observation made most consistently and confidently by study informants is that the research done in other developed countries and in several emerging Asian economies has been getting better and more abundant. In their view, improved capacity overseas, more than changes in what Americans have been doing, likely accounts for the increased share of S&E papers from foreign institutions. In an expanding literature, they see a continuing, even growing, American presence, but more marked growth occurring in other countries.

Advances in communication have made the international scientific literature more accessible to researchers in other countries. In this regard, advances in electronic communication loom large. As potential contributors to the literature, researchers can take advantage of improved electronic communication to collaborate more easily with distant colleagues and submit papers online. As readers, they can receive papers from colleagues via e-mail, find information in electronic archives and databases, and access scientific communications that cannot be found in a local university library. But, beyond electronic communication, increased worldwide capacity to communicate in a common scientific language, English, has also played a role.

As the largest and most influential producer of scientific articles in the world and a nation whose native language is also the dominant language of science, the United States was already at the center of the worldwide system of scientific communication before these advances occurred. Thus, journals were already highly accessible to U.S. researchers, both as contributors and as readers, at the outset of the period studied. Improvements in communication may have had a greater effect on the ability of researchers elsewhere in the world—especially those in nations or at institutions that were not prominent in research in the late 1980s—to keep up with their fields, produce research of a reasonable quality, and report their research in journals with a wider audience and a greater impact.

Institutional differences may also be affecting article counts. U.S. researchers perceive their universities and funding agencies as less attuned to quantitative measures of output and impact than their institutional counterparts in other countries. As a result, U.S. researchers may be less concerned with producing scholarly output in ways that score well on these measures. Study informants stressed the role of expert judgment in maintaining the commitment to quality in the U.S. system and saw the system as somewhat more oriented toward quality than in the past. They portrayed pressures toward scientific productivity as increasing, if anything, but as directed more toward enhanced quality than toward greater quantities of output. In contrast, many U.S. researchers interviewed saw other countries' efforts to improve research during the period under study as being increasingly driven by quantitative measures.

The study's findings provide little support for the idea that competing institutional demands are diverting faculty from research and publication. For the most part, informants said that neither teaching nor commercial activities were absorbing time that in the past would have been devoted to research and writing. Although some saw increased university concern about good teaching, and all agreed that institutional support for commercial activity was growing, faculty continued to believe that research was clearly the institutional concern that mattered most in shaping their behavior. It is possible, of course, that activities that compete with research for faculty time and attention, especially commercialization-related activities, have adverse effects on publication outputs that researchers themselves do not fully appreciate.

One of the most striking recent changes in how research is done has been the movement toward more collaborative work, and especially toward interdisciplinary and interinstitutional collaboration. Study data suggest that this trend can have either of two opposite effects on publication output. Insofar as it involves "complementary" collaborations that increase research output via a more rational division of labor, it should generate increased numbers of publications. However, insofar as the trend involves "integrative" collaborations that require extensive communication to synthesize different perspectives on a problem into a coherent piece of research, more people, money, and time may be required to produce a publishable article. It is possible that growth in publication output has slowed as a result of a movement toward integrative collaborations.

In the absence of solid data distinguishing the two kinds of collaboration, hypothesized relationships that turn on this distinction will be difficult to test. Yet, to interpret the significance in changes in publication output, it may be important to use relatively sophisticated understandings of the interplay among article counts, impact, collaboration, and interdisciplinarity. Some study informants suggested that successful integrative collaborations have had disproportionate impact on their fields and that the United States has been in the forefront of movement toward this type of collaboration. If U.S. researchers, compared with researchers in other countries, have been more rapidly increasing their investment of time and resources in this type of collaboration, this might affect article counts.

Another possibility, suggested by a reviewer commenting on an earlier draft of this report, is that the United States may experience "a productivity price associated with being in the leading position" in research worldwide. Insofar as research leadership requires personnel with the technical and interpersonal skills to forge successful integrative collaborations, such personnel may cost even more than personnel who are "merely" in the forefront of their specialties. Similarly, insofar as research leadership requires innovative use or development of state-of-the-art equipment, equipment costs may be even greater than for techniques that "merely" need the most advanced equipment available. Moreover, research in the "leading position" is more susceptible to failures and false starts that do not generate publications. To be sure, these costs are borne by leading research groups around the world; if they occur more often in the United States, they certainly do not occur exclusively in this country.

The effects of time and effort devoted to securing funding may also warrant further consideration. Some researchers said that in the United States competitive mechanisms intended to spur productivity sometimes and increasingly had had the opposite effect, even as other countries profited from introducing more such mechanisms. Similarly, although the study team heard indications that regulatory and governance burdens were increasing at U.S. universities, there was little to indicate whether the situation in other countries was different in this regard.