Glad to have been part of this project. I think the results here shed light on the intricacies of replication and how we can improve the process. Thanks for all the hard work @breznaunate.bsky.social, Eike Mark Rinke, and @kunkakom.bsky.social!
Reposted from
Nate Breznau
We crowdsourced 85 teams to indep. computationally reproduce results from a single study in randomized conditions. Reproducibility:
• depends heavily on the transparency of materials
• high (~95%) for same sign/sig, but low (btw. 48-77% depending on transparency) when requiring within 0.1 effects
• depends heavily on the transparency of materials
• high (~95%) for same sign/sig, but low (btw. 48-77% depending on transparency) when requiring within 0.1 effects
Comments