Towards computational reproducibility: researcher perspectives on the use and sharing of software

Research software, which includes both the source code and executables used as part of the research process, presents a significant challenge for efforts aimed at ensuring reproducibility. In order to inform such efforts, we conducted a survey to better understand the characteristics of research software as well as how it is created, used, and shared by researchers. Based on the responses of 215 participants, representing a range of research disciplines, we found that researchers create, use, and share software in a wide variety of forms for a wide variety of purposes, including data collection, data analysis, data visualization, data cleaning and organization, and automation. More participants indicated that they use open source software than commercial software. While a relatively small number of programming languages (e.g. Python, R, JavaScript, C++, Matlab) are used by a large number, there is a long tail of languages used by relatively few. Between group comparisons revealed that significantly more participants from computer science write source code and create executables than participants from other disciplines. Group comparisons related to knowledge of best practices related to software creation or sharing were not significant. While many participants indicated that they draw a distinction between the sharing and preservation of software, related practices and perceptions were often not aligned with those of the broader scholarly communications community.