Daniel Hart Baker

New #preprint about our experiences offering 'reproducibility as a service' for a six month period: https://psyarxiv.com/k8d4u/

We made 10 papers #reproducible from @YorkPsychology covering a wide range of study designs, supported by funding from Research England 1/6

Aug 31, 2023 at 4:57pmWeb
Show threadDaniel Hart BakerAug 31, 2023
Using #RMarkdown we converted standard manuscripts into a fully reproducible format, with all analyses computed from the raw data and results input directly into the text. We learnt a lot along the way, and developed some new methods, including using Docker and Github Actions 2/6
Show threadDaniel Hart BakerAug 31, 2023
All of our contributing authors were extremely cooperative, and clearly saw the benefit of their work being reproducible. We think this is how all scientific results should be reported. However we recognise several barriers to wider adoption, chiefly expertise and time 3/6
Show threadDaniel Hart BakerAug 31, 2023

We have made available some repositories of example markdown scripts, as well as a ‘cookbook’ of useful tips here:

https://github.com/bakerdh/ReproduceMe

Hopefully this will help others to develop the necessary skills 4/6

GitHub - bakerdh/ReproduceMe: Handbook for computational reproducibility project

Handbook for computational reproducibility project - GitHub - bakerdh/ReproduceMe: Handbook for computational reproducibility project

GitHub
Show threadDaniel Hart BakerAug 31, 2023
However if anyone is interested in arranging a bespoke workshop to learn our process, or commissioning work along these lines, please do get in touch by DM 5/6
Show threadDaniel Hart BakerAug 31, 2023
Team members included the fantastic @KHansford Bartholomew Quinn, Erin Warden-English, Kiki Berg and Federico Segala. We have submitted the preprint to Meta-Psychology and open reviews are very welcome (details in preprint) 6/6
Show threadDaniel Hart BakerSep 6, 2024

The final version of this paper has now been published:

https://doi.org/10.15626/MP.2023.4021

LnuOpen | Meta-Psychology

Show threadDeborah ApthorpAug 31, 2023
@bakerdh what a great idea! Universities should pay for this sort of thing at scale.
Show threadDaniel Hart BakerSep 1, 2023
@deboraha Couldn’t agree more!
Show threadDigital Research AcademySep 14, 2023

@bakerdh @KHansford that sounds like a very fine idea 👏

Get in touch if you or any of your collaborators want to join our trainer network 🤝

https://digital-research.academy/contact/

Contact

Quality Training for Quality Research

DIGITAL RESEARCH ACADEMY
Show threadSteve LindsaySep 2, 2023
@bakerdh @YorkPsychology Have not read the paper but like the idea. I hope you are documenting your interactions.
Show threadDorien HuijserSep 2, 2023
@bakerdh What a cool project, thanks for sharing your findings! I have a few questions that I did not find the answers to in the preprint:
- Did you do this pilot full-time for 6 months? How many members did your team have? It sounds quite intensive!
- Did you only use studies with open data, what if the data cannot be made publicly available, e.g., due to privacy reasons?
- What did your local support staff think of this pilot (e.g. university library), is it an option to collaborate with them?
Show threadDaniel Hart BakerSep 3, 2023

@dorienhuijser

Hi Dorien,

It wasn't full time for any of us, and there were six people involved (the authors of the preprint).

Yes, that's a definite restriction, as it's unlikely we could do all analyses from the raw data if there were privacy concerns (i.e. with structural MRI data).

We did talk to the library open research people, who were supportive, but there wasn't anything we particularly needed from them.

Thanks for your interest!