Welcome to the RobustiPy homepage!

RobustiPy is an efficient multiversal library with model selection, averaging, resampling and out-of-sample analysis. It analyses various output spaces, in addition to the control variable space (e.g. multiple dependent variables, estimands of interest, etc). Developed for Python and R, it is designed to be both accessible and computationally efficient.

Note: This site (and project!) are a work in progress!

Installation

Installation is as simple (in Python) as:

pip install robustipy

and – potentially in the future for R – as:

install.packages("robustipy")

Examples: A series of toy examples with which might be useful to get you up and running can be found here. A series of empirical examples which might be useful can be found here.

Hackathon: We are holding a hackathon regarding RobustiPy in Oxford during June 2024. Participants can click here for further information on the day’s activities!

Inspiration: We found great inspiration in the following three papers, which you might also enjoy reading:

1. Simonsohn, U., Simmons, J. P., & Nelson, L. D. (2020). Specification curve analysis. Nature Human Behaviour, 4(11), 1208-1214.
2. Young, C., & Holsteen, K. (2017). Model uncertainty and robustness: A computational framework for multimodel analysis. Sociological Methods & Research, 46(1), 3-40.
3. Gelman, A., & Loken, E. (2013). The garden of forking paths: Why multiple comparisons can be a problem, even when there is no “fishing expedition” or “p-hacking” and the research hypothesis was posited ahead of time. Department of Statistics, Columbia University, 348(1-17), 3.

Contributing: We welcome contributions (or matters arising, suggestions for features, or bug alerts!) from anybody and everybody interested in our work! Please raise all issues on GitHub.

License: This work is made available under a GNU General Public License v3.0.

Who: RobustiPy is developed and maintained by Daniel Valdenegro Ibarra, Charles Rahal, and Jiani Yan at the Leverhulme Centre for Demographic Science and the Centre for Care. Comms and administrative support gratefully recieved from Bradley Hall-Smith, Hannah Calkin and Dan Williamson.

Thanks!: We’re grateful to comments from participants at an LCDS Internal Meeting, the Centre for Care, IC2S2 (2023), PAA (2024), and the PSI Software Engineering Group. Funding gratefully received funding from the ESRC (Centre for Care), and the Leverhulme Trust (Grant RC-2018-003), the Leverhulme Centre for Demographic Science (LCDS), and Nuffield College.