This working paper explores trends from 1988 to 2011 to see how the number of academic scientific publications (output) is related to funding for academic R&D in S&E and the number of U.S.-trained, doctorate-level academic researchers in science, engineering, or health (inputs). Trends are reported at aggregate and field levels. By using the 1994 Carnegie classification of institutions of higher education,[3] the paper also examines trends for different types of universities and colleges: public versus private and universities with the highest level of research activity (Research I universities) versus less research-extensive universities and colleges.

The analysis used three sources of data:

• The National Science Foundation's (NSF's) annual Survey of Research and Development Expenditures at Universities and Colleges for estimates of U.S. academic R&D expenditures from FY 1986 to FY 2009. (The survey was renamed the Higher Education Research and Development [HERD] Survey in FY 2010).
• NSF's Survey of Doctorate Recipients (SDR) for estimates of the number of U.S.-trained doctorate-level academic researchers for whom research was a primary or secondary work responsibility from 1993 to 2010.[4]
• Thomson Scientific's annual Science Citation Index and Social Sciences Citation Index for data on the number of highly cited U.S. academic publications from 1988 to 2011.

Notes

[3] The Carnegie classification of institutions of higher education is widely used in higher education research to characterize and control for differences among academic institutions. The 1994 classification includes six broad categories of institutions, including doctorate-granting research universities, master's colleges and universities, baccalaureate colleges, associate's colleges, specialized institutions, and tribal colleges and universities. Among doctorate-granting research universities, institutions with the highest level of research activity are designated as Research I universities in the 1994 classification.

[4] Because of changes in the survey questionnaire about work activities, data prior to 1993 are not directly comparable to data from 1993 and afterward. As a result, this analysis draws data from the SDR in the nine survey cycles spanning 1993–2010 (data for intervening years are interpolated by applying the annual growth rate computed using the two surrounding years to the missing years).