University of California, Berkeley
To develop and promulgate best practices in the review of statistical research software
Supported by the Sloan Foundation since 2013, rOpenSci is an open source community that develops research software inside the R computing environment, especially focusing on the creation of expansions and modifications of R useful to the research scientist. The rOpenSci community has become known for high-quality, trusted research software, largely because every user-developed package is run through a robust peer review process before it is added to the rOpenSci suite. Increased access to basic data science skills combined with the demand for research software has led to rapid growth in software packages, many of which implement statistical methods. A large proportion of these software packages are highly variable in quality and lack appropriate tests to ensure that the software produces correct results consistently, across a variety of conditions. Much of this is due to the lack of clear standards (within and across fields) and guidance on how to implement them. Funds from this grant support a two-year effort to address this pressure on two fronts: to extend the rOpenSci model of scientific software peer review into substantial assessment of the implementation of statistical methods, and to build out a technical infrastructure to manage this expanded review process.