breakthrough innovation needs to be better understood, and more useful data on federal investment in R&D would help.
While sympathetic to the need for more information to illuminate the trends in research and development and the federal role in that enterprise, the panel is also sensitive to the concern that federal agencies not be over-burdened with requirements for additional data reporting. A principal reason the current data are less reliable than desirable is that, in the view of many agencies, filling out NSF reports is labor-intensive and difficult and the benefits barely justify the high cost in labor and other resources (Touhy, 1998). Rather than simply increasing the reporting burden, it would be preferable to consider new data search and analysis technologies tied to expanding efforts to make government data accessible (such as http://www.data.gov), which might gradually make it easier to obtain the raw data for the R&D surveys in ways less burdensome than the current individually conducted compilation processes undertaken separately at each agency.
With these user needs and challenges in mind, the panel set about to identify a step-by-step process that would lead to improvement in the federal data on R&D spending. Following this introduction, Chapter 2 provides a description and critique of the current status of the two surveys that now provide the information used to portray federal R&D spending. Chapter 3 focuses on current problems and makes suggestions for a few relatively modest improvements that could be made in the short term (the next four years or so) to the current system of surveys. We then urge NSF to focus attention on a medium-term solution (over roughly 4 to 10 years) that would make use of new technologies and maturing automated databases and set the stage for long-term changes in the collection system—beyond 10 years. Chapter 4 describes the potential and limitations of the use of administrative data for collecting and compiling information on federal R&D spending and identifies opportunities for transitioning to a new system of data collection.
In Chapter 5, we explore some cutting-edge possibilities for long-term changes in the way in which R&D is viewed and the manner in which information about it is collected. In Chapter 6, these threads are gathered into a recommended course of action, which would take NSF through the process of making small short-term improvements in the surveys, undertaking an initiative to build a much fuller, more useful administrative records–based system, and laying the basis for even more revolutionary changes in the long term.