Re:U.S. regulators accuse Palantir of bias against Asians
by kafkaesq on Monday September 26, 2016 @05:14PM
(#12585618)
It's an elementary statistics calculation (which I'm not saying was done right; but that's the direction it was most likely coming from).
Left unstated was what
P(portion of Asian applications being >= 74%)
should be given that, if I'm not mistaken, the percentage of Asian undergraduates (even at upper-tier schools) seems to top out at around 45% (a very rough figure based on quick eyeballing; but I doubt it pushes 60%).
Re:U.S. regulators accuse Palantir of bias against Asians
by malchow on Monday September 26, 2016 @05:48PM
(#12585893)
I understand that, thanks. The problem of course is that it bears no relation to actual reality. It may be the case that, for a pipefitter, every human body with Pipefitter Certification 409B can do the job identically to his fellow 409Ber. But there are very few jobs like that–trending to zero.
Result of this case: lawyers make a few million bucks. Government gets a few million bucks. Palantir has a few million bucks fewer with which to grow and produce new value. Fewer people get hired. No one wins. Administrative overreach: it's a thing.[1]
[1] https://www.amazon.com/Administrative-Law-Unlawful-Philip-Ha...
Re:U.S. regulators accuse Palantir of bias against Asians
by zachsnow on Tuesday September 27, 2016 @12:01AM
(#12587357)
Not sure about pipe fitting, but for welding, say, or framing? There are definitely 10X welders. So your "trending to zero" point resonates with me (being a shitty welder).
Re:U.S. regulators accuse Palantir of bias against Asians
by nicolashahn on Monday September 26, 2016 @05:21PM
(#12585675)
Without actually doing the math, I believe it means that the odds that, if 21 names were pulled out of a hat with 130 names of "qualified applicants," and 73% were Asian, the probability that only 4 names (or less, I assume) are Asian is 1:1,000,000,000.
Re:U.S. regulators accuse Palantir of bias against Asians
by barkbro on Monday September 26, 2016 @05:55PM
(#12585925)
Isn't this correct though? That the probability of picking only four asians (if you chose at random) is just over one in a hundred million
Re:U.S. regulators accuse Palantir of bias against Asians
by sib on Monday September 26, 2016 @08:03PM
(#12586647)
Yes - but the objection is that no one in his/her right mind should be making hiring decisions at random. So, comparing it to choosing at random in rather obtuse (and clearly intended only to score cheap points by the person who filed the lawsuit.)
What in the world does that alleged statistic mean?