This INSPIRE award is partially funded by the Science of Science and Innovation Policy Program in the Division of SBE Multidisciplinary Activities in the Directorate for Social, Behavioral, and Economic Sciences and the NSF Office of Cyberinfrastructure.
This project undertakes an interdisciplinary and novel approach to the problem of measuring the effects of investment in cyberinfrastructure to universities' production processes of research outputs and vital educational services. A decision to support funding of the infrastructure that supports research, or a decision to support funding of focused research activities, is an increasingly critical decision with far-reaching impacts not only to the institutions receiving those funds, but also to national competitiveness. While it is generally agreed that cyberinfrastructure is essential to scholarly inquiry in some science fields, the scope of cyberinfrastructure's broad effects on the growth of knowledge, to the academic enterprise, and to areas of science has not been explicitly quantified.
This work extends the state of knowledge of frontier efficiency analysis (FEA) techniques, a rigorous statistically-grounded approach, and uses FEA in a novel application to examine the returns to cyberinfrastructure investments in research institutions. Such work is strongly interdisciplinary: experts in econometrics will be working collaboratively with experts in computing and cyberinfrastructure to apply and extend the state-of-the-art in data management, extend our developed Unified Data Framework, and prepare and curate a significant new body of data of great interest to decision-makers in cyberinfrastructure investment and science policy. This research also makes a contribution to statistical estimation theory, developing new central limit theorem results applicable to means of nonparametric frontier efficiency estimators, permitting testing of hypotheses regarding returns to scale and other aspects of universities' productivity. To overcome the problem that nonparametric frontier efficiency estimators are biased in finite samples and have slow convergence rates that depend on dimensionality of the particular problem, this project uses subsampling ideas to construct new statistics for hypothesis testing in nonparametric frontier efficiency estimators for which central limit theorem results can be obtained.
Finally, the project makes a dual contribution to science and innovation policy and to cyberinfrastructure investment decisions, using frontier efficiency analysis (FEA) techniques to evaluate the productivity (i.e., efficiency in transforming inputs into outputs) of research institutions who have received various amount of cyberinfrastructure investments. Project personnel will collect and curate a large body of data on educational and research institutions' productivity and performance as a part of this analysis. The analysis of this data will allow examination of important science-policy questions on investment in cyberinfrastructure, such as "Does cyberinfrastructure enhance universities' research and educational output? If so, how, where, and by how much? Are returns to scale in cyberinfrastructure investment increasing, constant, or decreasing? What might be a method of allocating future scarce cyberinfrastructure resources efficiently across universities?" The project will support 50% of a postdoctoral associate and a full-time Ph.D. student in interdisciplinary econometric and computer-science research. The techniques and tools utilized in this research are to be incorporated into a newly-developed course in Data-Enabled Science, which, over the course of the project, is expected to reach over 40 students in a variety of science and engineering disciplines.
The overall project is potentially transformative to the science of science policy: the project's development and demonstration of the use of formal measures for capturing the effectiveness of investment has the potential to significantly change the processes used to construct portfolios of funded scientific-research projects, and to add significant quantitative support to the policies by which these decisions are made. The specific case study (the effect of cyberinfrastructure investments on research and educational outcomes) can immediately inform institutions, states, and governmental funding bodies about opportunities for high-priority future investments in this area, particularly by identifying overlooked opportunities where modest investments will yield significant returns.
