Project selection in NIH: A natural experiment from ARRA

Using a natural experiment in research funding by the National Institutes of Health (NIH) following the American Recovery and Reinvestment Act (ARRA) of 2009, we study the NIH's revealed preference in project selection. We do so by comparing the characteristics of the projects additionally selected for funding due to an unexpected increase in resources under the ARRA with those supported through regular NIH budget. We find that the regular-funded projects are on average of higher quality, as measured by the number of publications per project and the impact of these publications, than ARRA-funded projects. Moreover, compared to ARRA projects, regular projects are more likely to produce highest-impact articles and exhibit greater variance in research output. The output from regular projects also seems more closely fitting the purpose of funding. The differences in project quality are largely explained by observable attributes of the projects and research teams, suggesting that the NIH may use these attributes as cues for discerning underlying project quality. In addition, ARRA projects are more likely than regular projects to involve investigators with past grant experience. Many of these inter-group differences are specific to RO1 grants, the largest funding category in the NIH. Overall, these results suggest that the NIH's project selection appears generally in line with its purported mission. In particular, our results contrast starkly with the frequent criticism that the NIH is extremely risk-averse and unwarrantedly favors experienced investigators. We discuss the implications of our findings on the NIH's behavior in project selection. (C) 2015 Elsevier B.V. All rights reserved.