WIDE Data Analysis: enabling wide, detailed and interactive data practice
In data analysis there is a tension between perceived methodological rigour and analytic richness. While formally pre-stated analysis protocols may confer ‘rigour’, through repeatable (and automatable) plans, in locking-down the data analysis process they may risk important phenomena being missed, or alternative perspectives being overlooked. We present WIDE Data Analysis to improve the balance with which data analysis is conducted. Our proposal consists of four strategies – [w]idening, [i]nteracting, [d]etailing and [e]nabling – and accompanying qualities or values that a successful data analysis should attain – replication value, didactic value, generative value, ludic/play value. We explain how critical data practices such as sensitivity analysis, many-model thinking, visual data analysis and storytelling underpin these strategies and values. Inspired by D’Ignazio and Klein’s data feminism (2020), we then develop prompting questions that compel analysts to reflect on the key decision points of an analysis, and envisage how these prompting questions, together with the critical data practices, might be called upon in AI-backed computational notebook environments. Our thesis is that at a moment when large parts of data analysis are being automated, frameworks such as WIDE that eschew a recipe-based approach to science are urgently needed.
Authors
- Rachel S. Franklin (ORCID: https://orcid.org/0000-0002-2614-4665)
- Levi John Wolf (ORCID: https://orcid.org/0000-0003-0274-599X)
- Çağatay Turkay (ORCID: https://orcid.org/0000-0001-6788-251X)
- Jessica R. Hullman (ORCID: https://orcid.org/0000-0001-6826-3550)
- Kai Xu (ORCID: https://orcid.org/0000-0003-2242-5440)
- Andrew Gelman (ORCID: https://orcid.org/0000-0002-6975-2601)
- María-Cruz Villa-Uriol (ORCID: https://orcid.org/0000-0002-3345-539X)
- Roger Beecham (ORCID: https://orcid.org/0000-0001-8563-7251)
- Edwin Pos
- Dianne Cook
- Benjamin Bach
- Hadley Wickham
- Jo Wood
- Daniel Archambault
Institutions
- University of Washington (US)
Publication Details
- Journal
- SocArXiv (OSF Preprints)
- Published
- 2026-11-20
- DOI
- https://doi.org/10.31235/osf.io/4qesy_v1
- Primary Topic
- Scientific Computing and Data Management
- Type
- preprint