Has “double-dipping” cost U.S. science funding agencies tens of millions of dollars?
Last year, an audit by the U.S. Government Accountability Office found “a potential for unnecessary duplication” among the billions of dollars in research grants funded by national agencies. Some researchers, it seemed, could be winning more than one grant to do the same research.
Prompted by that report, Virginia Tech’s Skip Garner and his colleagues used eTBLAST, which Garner invented, to review more than 630,000 grant applications submitted to the NIH, NSF, Department of Defense, Department of Energy, and Susan G. Komen for the Cure, “the largest charitable funder of breast cancer research in the United States.” The approach was not unlike those by publishers to identify potential article duplications.
In a Comment published today in Nature, they report that they found 1,300 applications above a “similarity score” cutoff of 0.8 for federal agencies, and 0.65 for Komen documents — “with 1 indicating identical text in two same-length documents, and more than 1 representing identical text in one piece that is longer than the other.”
When they manually reviewed those 1,300:
We found that 11% of the pairs with a similarity of more than 0.8 (or 0.65) had overlapping aims, hypotheses or goals. For these 167 pairs the total money involved was around $200 million (including both grants of the pair) over the entire time records are available. The average size of the first award was 1.9 times that of the potentially overlapping one, so an estimated $69 million of possible overlap funds were found.
The authors note a number of limitations of their analysis, notably that they did not have access to entire grants, which would have required Freedom of Information Act (FOIA) requests — and that data from some agencies was incomplete. (A Nature news story based on FOIA requests for about 22 of the potential duplications Garner’s team uncovered describes several cases in detail.) Still, the authors write:
Even if $200 million in duplicated grants represents the full extent of the problem, then some may argue that less than 0.1% of funding since 1985 is too small an amount to warrant concern. But that it is research money that cannot be used to fund the next scientific breakthrough.
If not the next breakthrough — given how rarely those actually occur — then the money could at least fund projects that fell below current funding lines. The authors call for a “central database of grant information from all agencies” that could be searched for potential duplicates.