|
Abstract
|
This paper reports the initial findings from large scale surveys of the scientists based in Japan and the US on the knowledge creation process in science from a comparative perspective. The survey in Japan was jointly conducted by the Institute of Innovation Research (IIR) of Hitotsubashi University and the National Institute of Science and Technology Policy (NISTEP) from the end of 2009 to the summer 2010 . The survey in the US was implemented by the Georgia Institute of Technology, in collaboration with IIR and NISTEP, from autumn 2010 to early 2011. It collected around 2,100 responses from scientists in Japan and 2,300 responses from scientists in the US on their research projects that generated the scientific papers subjected to the surveys. <br>Roughly one-third of the sample are from highly cited papers (top 1% in the world, H papers hereafter) in each science field and the rest are from randomly selected papers (N papers hereafter). We call the research projects that yielded H (N) papers H (N) projects. The population of the survey was articles and letters in the Web of Science database of Thomson Reuters. The response rate was 27% in Japan and 26% in the US. The survey covered all scientific fields, including social sciences. For our US-Japan comparison, we have adjusted the field composition differences of the two countries, using the field composition in the world as the baseline, since physical sciences have larger shares in Japan than in the US (the share of all physical sciences is 58% in Japan, compared with 47% in the US). The survey characterized the motivations of the research projects; the knowledge sources which inspired the projects; uncertainty in the knowledge creation process; research competition; composition of the research team; sources of research funding; the research outputs, including papers, patents, and licenses; and the profile of scientists.<br>Major findings are as follow:<br>1. More than 70% of the responding scientists belong to higher education institutions in both countries (73% of the H papers in Japan and 76% of the H papers in the US); 10% to 20% of the respondents belong to pubic research organizations (higher in Japan); and around 5% of the respondents belong to private firms in both countries.<br>2. Japanese respondents are younger: as for submission age, the average ages of respondents in the natural sciences are 42.8 (H papers) and 43.7 (N papers) in the Japanese sample; and 45.6 (H papers) and 46.7 (N papers) in the US sample. Around 90% of respondents (89% in Japan and 92% in the US) had doctoral degrees when the research was launched. Japanese respondents are as mobile across organizations as US respondents, controlling for age. <br>3. Pasteur’s quadrant (both “Pursuit of fundamental principles/understandings” and “Solving specific issues in real life” are very important motivations) occupies a significant part of scientific research in both countries. Among H projects, the share in Pasteur’s quadrant is more than twice as high in the US than in Japan (33% vs. 15%). <br>4. Bohr’s quadrant (only “Pursuit of fundamental principles/understandings” is very important) accounts for the largest share of research projects; 45% (35%) of the H (N) projects in Japan and 46% (42%) of H (N) projects in the US. Edison’s quadrant (only “Solving specific issues in real life” is very important) accounts for 15% (16%) of the H (N) projects in Japan and 11% (15%) of the H (N) projects in the US. <br>5. Research involves very substantial uncertainty in both countries. Both the main result of the paper and the research process were as initially expected or planned only for 11% of the H papers in Japan and 14% in the US (17% of the N papers in both countries). Research process uncertainty is high in Pasteur's and Bohr’s quadrants in both countries. <br>6. In both countries, the research output of the paper often found answers to questions not originally posed, that is, serendipity in the sense of (Stephan (2010)) occurred. H papers involve more serendipity and a serendipitous output is more often observed in a research project involving more process uncertainty in both countries. Thus, scientific research not only yields the results (sometimes more than expected) to the original questions but also those to the questions not originally posed. Appreciating such option value would be important for scientific research funding.<br>7. In both countries, most researchers recognize the extent of research competition ex-ante (only a minority chose “don’t know”). A significant number of researchers were concerned with priority loss (more than 50% of the researchers in Japan and 23% of them in the US for H papers). Such concern is stronger in H projects than in N projects. It increases with the number of competitors recognized ex-ante. Priority threat is seen as greater in Japan than in the US.<br>8. By far, the most important knowledge source for suggesting the research project is scientific literature in both countries. Colleagues in the organization (a university, a laboratory, etc.), visiting researchers or post-doctoral students in the organization and past research collaborators follow scientific literature in both countries. The locations of the important knowledge sources are often domestic (exceeding 60%) for the US scientists, while they are often abroad for Japanese scientists, except for the knowledge sources embodied in researchers and facilities.<br>9. Research is more actively managed in H projects than in N projects in both countries: ambitious goal setting, information sharing and discussions in a team, division of research tasks for outsourcing of a research task, improvement of facilities and program, and development of a research community <br>10. US scientists seem to make more use of research tool databases, and to engage remote researchers, using the internet, in their research projects.<br>11. Most scientific research is done by a team in both countries. The share of single authored papers is 3.0% in Japan and 5.4% in the US for H papers. The median author size is 6 in Japan and 5 in the US for the H papers (4 for N papers in both countries). A researcher who provides only materials or research facilities is often added as an author in both countries, and authorship is more expansive in Japan, which is consistent with a larger size of authors per paper in Japan. <br>12. Young scholars (students and postdoctoral fellows) are important contributors for research efforts in both countries. Post-doctoral students and doctoral students are often the first authors of H papers when the order of the authors is according to their contributions in both countries (young scholars account for 40% in Japan and 50% in the US in the case of higher educational institutions).<br>13. The involvement of young foreign-born scholars is important in both countries. It accounts for more than 70% of the first authors of H papers in the US and around one-third in Japan. <br>14. Research teams have more diversified memberships in terms of specialized academic fields, specialized skills, origins of birth and types of sectors in H papers than in N papers in both countries. The US teams are significantly more diversified in the origins of birth than the Japanese teams (80% of teams in the US involve researchers from more than one country vs. 50% in Japan for H papers). Given that international co-authorship in terms of the locations of affiliated organizations of the US is only modestly larger than that of Japan (24% in Japan and 29% in the US in 2005 – 2007), the above difference largely reflects the inflow of foreign-born scholars in the US.<br>15. The time-lag between the conception of the research project and its launch is mostly a year or less in both countries but has a longer tail in Japan. Time-lag between the launch of the project and the submission of the focal paper is shorter for H project than for N project and shorter in the US.<br>16. In terms of the median of the total labor input per project in natural sciences, the projects in Japan spend about 3 times as much as those in the US in both H and N projects. The median number of papers published per project is also roughly 3 times larger in Japan than in the US, suggesting that the concept of “project” is interpreted or defined more narrowly in the US than in Japan. In addition, the research projects in the US are significantly more money intensive than those in Japan. However, some of this difference is due to accounting practices in the two countries (e.g., the extent to which the grant includes all the direct and indirect costs of research).<br>17. H projects are not only large but significantly more money intensive (higher expenses relative to man-months) than N project in Japan. The median budget of H projects is 5.0 (1.8) times more than that of N projects, while the median size of man-months is only 1.4 (1.5) times larger in Japan (the US). <br>18. The majority of research projects of higher education institutions in Japan were funded by a combination of intramural and extramural sources. In contrast, more than 50% of research projects of US universities were funded only by external sources. On the other hand, in public research institutions, about a half of research projects in the US are conducted using only intramural fund, while only about one-sixth to a quarter of research projects in Japan are.<br>19. Mission-oriented programs account for a significantly larger share of the research funding in the US than in Japan (43% (22%) of the H projects on the simple average and 50% (38%) on the weighted average in the US (Japan)). Industry accounts for a relatively small and similar shares of funding in both countries (8% of the H projects in Japan vs. 9% of the H projects in the US in the simple average). Surprisingly, industry funds a greater share of the projects of higher educational institutions in Japan than in the US (5% of the H projects and 8% of the N projects in Japan vs. 3% of the H projects and 5% of the N projects in the US). If we measure industry funding by the percent of projects with at least some industry funding, this contrast is even greater (As for H projects, 24% of Japanese projects have at least some industry funding, compared to 12% of US projects).<br>20. The median number of refereed papers produced by H projects is 1.9 (1.7) times larger than that of N projects in Japan (the US), which is larger than the research labor input ratio but smaller than the research money ratio between H and N projects. The distribution of the number of refereed papers produced from a project is highly skewed (it has a long right tail). <br>21. Educational outputs of the research projects are also important, especially training of PhDs and postdoctoral fellows. More than 73% (59%) of H projects produced a PhD in Japan (the US). Educational outputs are larger in H projects than in N projects in both countries. The research projects also often produced materials and other research tools.<br>22. Research projects resulted in more patent applications in Japan than in the US (39% of the H projects and 22% in N projects in Japan. The corresponding shares are 16% and 8% in the US). They also resulted in more licensing or assignments of a research result in Japan (14% of the H projects and 7% of the N projects. The corresponding shares are 9% and 4%). Note, however, that the projects are more broadly defined in Japan (roughly 3 times more man-months and published papers). H papers are more often commercialized in both countries. There exist significant variations across science fields: materials science, chemistry and engineering are the most commercially active fields in both countries, while life sciences and clinical medicines are only moderately commercially active<br>23. A majority of licensing and assignment (70 to 80%) were associated with the provision of know-how in both countries, indicating the importance of technology transfer effort on the part of universities. <br>24. Only a relatively few research projects resulted in start-ups in both countries (2% of the H projects in Japan and 4% of the H projects in the US).<br>The paper also discusses the implications of these research findings on research on research and on science policy.<br>
|
