// the find
jtleek/datasharing
The Leek group guide to data sharing
A short guide from Jeff Leek's biostatistics group at Johns Hopkins explaining how to hand off data to a statistician — raw data, a tidy dataset, a codebook, and a processing script. Aimed at researchers and grad students in quantitative fields who have data to analyze but aren't statisticians themselves. It's essentially a one-page contract between scientist and analyst.
The four-deliverable framework (raw data, tidy data, codebook, script) is genuinely useful and worth memorizing. The explanation of censored vs. missing data is more careful than most guides at this level. The warning about Excel eating date/time values in CSVs is the kind of hard-won practical advice that saves real pain. 242k forks suggests it's been assigned in courses worldwide — the writing is clear enough to survive that.
It's a README with no versioning, no tooling, and no examples — the 'tidy data' section links out to a 2014 Hadley paper and a Vimeo video rather than showing a worked example inline. The codebook section recommends Word files, which is a strange choice for something meant to support reproducibility. Last updated 2024 but the content reads like 2013 and doesn't mention Quarto, DVC, or any modern data provenance tooling. There's no discussion of large datasets, databases, or anything beyond spreadsheets — the scope is narrow even for its intended audience.