From 1988 to 2006, the ratio of academic R&D spending to publications increased regardless of counting method used; however, this trend was not evident from 2007 to 2009. From 2009 to 2011, this ratio began to increase again. Although the increase was evident at all institution types, the rate of growth was highest for less research-extensive universities and for public schools. Private schools, especially those that are Research I institutions, showed very little or no growth in the spending-to-publications ratio from 1991 to 1996.
Using whole counts, the ratio of articles to academic researchers increased during the first two-thirds of the 2000s. However, from 2007 to 2011, this ratio gradually declined to the 1994 level. When publications are counted fractionally, there was a slight decrease in the average annual per capita output of publications from 1994 to 2011. Neither estimate takes into consideration academic researchers who received their doctorate overseas. Finally, it appears that the relative flattening of annual academic S&E publication output during the 1990s and early 2000s has ended (figure 1).
This analysis focused on academic researchers whose primary or whose primary or secondary work responsibility was R&D. Further study focused more narrowly on primary academic researchers would be beneficial. For example, a more detailed analysis of the work activities over time of primary academic researchers at private and at public universities and colleges could shed light on the extent to which these researchers are being called upon to take on more administrative or other responsibilities that do not result in publications.
Throughout the quarter century covered in this analysis, many factors have contributed to trends in academic S&E publication output. As this paper notes, one of the most important factors is the increasingly collaborative nature of academic R&D. This growing collaboration is reflected by downward trends in the number of articles counted fractionally per researcher, and analyses focusing on finding ways to measure the relative contribution of various institutions or individuals to articles in specific S&E fields could shed more light on this phenomenon.
Further analysis of changes in the research process, both by and across fields, could explore and potentially begin to measure the extent to which academic R&D is becoming more complex, involving time-consuming integration of diverse perspectives to address multifaceted research problems. For example, it would be helpful to analyze what fields are most prominently represented in interdisciplinary research and the impact this could have on spending per publication. Studies could explore ways to quantify the movement over the past two decades toward more integrative collaborations and how this has affected academic S&E publication patterns.
More research might clarify how and to what extent administrative responsibilities associated with academic R&D affect academic publications output. For example, research could measure whether and how much the life sciences are disproportionately affected by growing administrative responsibilities. Differences in research practices between public and private universities and colleges and between the nation's Research I institutions and others are also potentially fruitful topics. Finally, additional analysis could focus on the effect of changes in publication practices, by field and at the academic sector level.
 Beginning in 1995 (and periodically thereafter), the SDR has collected limited information on publication and patenting activity among respondents. To improve understanding of the products of research among U.S.-trained science, engineering, and health doctorates, NCSES has undertaken a project to link the SDR data files to several major publication and patent databases. This will provide direct measures of publication activities of SDR respondents and enable further analysis of factors related to publication and patent outputs. Data files will be available by license.