2 ms·
The paper is here: https://arxiv.org/pdf/2605.27371 https://arxiv.org/pdf/2605.27371 They find "disparate impact" of pymetrics across racial groups, but it doe
by verteu 3mo ago
The paper is here: https://arxiv.org/pdf/2605.27371 https://arxiv.org/pdf/2605.27371
They find "disparate impact" of pymetrics across racial groups, but it doesn't seem like they controlled for anything.
- efavdb 3mo agoThey also say that if they do the analysis globally the effect goes away. Curious, does that not imply that if one domain is biased against some group there would be another where the bias was in its favor?
- zeroonetwothree 3mo agoAlso has issues of random chance causing these differences. How many different positions are there that have the chance of a 80% effect?
- verteu 3mo agoThey're using a Benjamini–Hochberg correction (alpha=0.05) to account for multiple comparisons (see Table 2).