4 ms·
But regardless of whether the outcome fits your priors or not, you may wish to know whether the study has decent evidence, or is just noise. And GP has a good
by iguy 6y ago
But regardless of whether the outcome fits your priors or not, you may wish to know whether the study has decent evidence, or is just noise.
And GP has a good point. The authors have some automated scheme for grading how well-qualified people are. The recruiters also have some such scheme. If these two schemes differ, and the distributions aren't identical between the groups being compared, then you will detect group differences like what they see. To label these differences bias in the recruiter's process, you must be confident that you have less bias in your process. The details of this correction are going to be crucial.
- monoideism 6y agoYes, it's a good point. I made a snap response, and was wrong to do so. I just get tired of the ideological nonsense I see so often propogated by a certain group in the tech industry.
- thu2111 6y agoYes, but the list of criteria they use for controls is pretty reasonable and more importantly, unbiased by woke ideology. We all know many firms have a very strong bias towards women that grades them as more well qualified (desirable to hire), so we'd expect a re-run using purely objective metrics to detect that. The outcome of this study isn't a surprise, the only surprise is that the bias is "only" 10%.