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

Area of Science:

• Computer and Information Science
• General Research Methodology

Background:

• Research software, including source code and executables, poses challenges for scientific reproducibility.
• Understanding how researchers create, use, and share software is crucial for improving reproducibility.

Purpose of the Study:

• To survey researchers on their software creation, usage, and sharing practices.
• To identify characteristics of research software across various disciplines.

Main Methods:

• A survey was distributed to 215 participants from diverse research disciplines.
• Data collected included software forms, purposes, programming languages, and sharing/preservation perceptions.

Main Results:

• Researchers utilize software for diverse tasks like data collection, analysis, visualization, and automation.
• Open source software is more prevalent than commercial software among participants.
• While computer science researchers more frequently write source code, best practice knowledge is similar across disciplines.
• Discrepancies exist between perceived and actual software sharing and preservation practices compared to scholarly communication norms.

Conclusions:

• Research software is integral to scientific workflows but its diverse nature complicates reproducibility.
• Current software sharing and preservation practices require better alignment with scholarly communication standards.
• Further investigation into aligning researcher practices with reproducibility goals is warranted.