Staff project based in the School of Anthropology.
Spring 2016 round
A research result is reproducible when it can be re-derived, by the original authors and by others working independently. Yet there is mounting evidence that, across disciplines, many results cannot be reproduced as published. Often, this is because the approach to computing used by the authors (point-and-click, cut-and-paste) is itself a barrier to good practice for reproducibility.
This project aims to lay the groundwork for a culture of research reproducibility at Oxford, focusing on training in the effective use of computational tools in research. These tools are widely used in some disciplines, and they can enable researchers to easily track the process leading from data to results, so that it is fully reproducible. However, researchers often lack the opportunities, incentives and confidence to make best use of these tools.
To achieve this aim, we propose to set up a partnership with Software/Data Carpentry over the next academic year. Software/Data Carpentry are non-profit volunteer organisations focused on teaching researchers across disciplines the computing and data skills they need for effective and reproducible research. Successful instances of their workshops have been run by individual groups in Oxford, and related training is provided by units across the University (e.g. some Centres for Doctoral Training and the Digital Humanities Summer School). The partnership would allow researchers to coordinate activity and consolidate provision across divisions and departments. Partner institutions receive access to instructor training and guidance with capacity-building, in the process developing communities of practice around effective data-driven reproducible research. Concretely, the partnership will result in the training of six instructors based at Oxford and delivery of a minimum of four workshops open to students and staff across the University.