"Why don't we share data and code?" Nice and concise infographic about perceived barriers and benefits. Full article: https://royalsocietypublishing.org/doi/10.1098/rspb.2022.1113
@Flor__Mu
Alt Text:
In the middle has data and code sharing, perceived barriers and solutions.
One section has Scooping, you know your data ideas are plentiful, open data = more citations.
Lack of time, sharing data saves time, create a data management plan.
Lack of incentives, open data = more citations, scientific community recognition.
@Flor__Mu
Alt text:
Second section has insecurity, share with trusted collegaues recognize no "perfect code", emphasize growth and learning.
Data too large, split data into smaller chunks, share properties of data, advocate for storage funding. Complex workflow, write a detailed readme, use graphics to explain, automate where possible.
Unclear process, check with your library, many resources exist, check data templates.
@Flor__Mu
Alt text:
Unclear value, value is subjective, perspectives are limitless, opportunities for synthesis.
Sharing rights, form an agreement, check your library for resources, follow authors guidelines. Transient storage, avoid proprietary formats, share as soon as possible, use stable repositories. Sensitive content, aggregate and anonymize, provide sample data, generate synthetic datasets. Inappropriate use, write detailed metadata, be willing to help, set data governance plans.
@Flor__Mu the latest argument I’m getting against data sharing is that others will interpret the data differently. Like, ‘we won’t give you the data because we don’t agree with how you want to analyze it and what you think that analysis means’. Any suggestions for that one?
@paulralph I've been hearing that one too! Personnally, when hearing this argument, I usually suggest a detailed README, but i'd be happy to hear other suggestions.
I think the authors of the infographic have it covered under 'inapropriate use' (yellow part of the chart).
@Flor__Mu IMO data should be shared by default, and refusing to share should require a compelling justification, typically based on protecting privacy or trade secrets. Authors shouldn’t have the right to dictate how future users might interpret or use datasets.
@paulralph Agree with you. On a side note, I've just come accross this page with the notion of dual use, which I think raises interesting and valid questions too: https://www.recherche.uliege.be/cms/c_11374378/en/dual-use
@Flor__Mu during my term as Editor for @ASNAmNat we shifted to requiring data and code archiving (when data and code are used), along with a Data Editor group to check that the archives are complete and usable. I highly encourage other journals to follow suit. It was shocking to see how often archives were incomplete or not usable.
@danielbolnick @Flor__Mu @ASNAmNat It's been kind of wild to look at some of the shoddy "archived data and code" (or "oh I have to email the author") of other studies after spending a *lot* of time making my code + data clear. It's great to see journals starting to take that more seriously.
@analyst42 ⬆️