3 ms·
danso linked a different article than the one you discuss. The article they link discusses the analysis in greater detail, rather than telling stories. (which
by maxerickson 10y ago
danso linked a different article than the one you discuss. The article they link discusses the analysis in greater detail, rather than telling stories.
(which I'm not trying to weigh in on the quality of the main article, just pointing out that they did publish a discussion of their statistics. I can understand the editorial decision to break it into a separate article.)
- yummyfajitas 10y agoIt's the same underlying analysis. Their github repo contains the same script I linked to: https://github.com/propublica/compas-analysis https://github.com/propublica/compas-analysis https://github.com/propublica/compas-analysis/blob/master/Compas%20Analysis.ipynb https://github.com/propublica/compas-analysis/blob/master/Co... It even contains the same weasel words "almost statistically significant" for one analysis.
- maxerickson 10y agoYes, the statistical discussion is linked from the main article in a couple different places.