@SteveBellovin@infosec.exchange
Post #3630079
2026-07-06 14:47 UTC
Replies (3)
-
@rrb@infosec.exchange 2026-07-06 14:50
@SteveBellovin@infosec.exchange @spaf@mstdn.social I think that is a feature, not a bug. I have also heard from federal employees about grants being de-funded for including the word "bias," without considering that the word was being used describing statistics.
-
@arclight@oldbytes.space 2026-07-06 15:19
@SteveBellovin@infosec.exchange @spaf@mstdn.social Historically, radiation protection standards and data were based on the effects on adult white men. There are myriad examples of this kind of invisible bias.
-
@karlauerbach@sfba.social 2026-07-06 17:02
@SteveBellovin@infosec.exchange @spaf@mstdn.social Oy, the two great griefs of using statistical methods: The grief of defining the sample and then the grief of actually collecting it. On that second point, my mother and grandmother both worked for Princeton Research - Gallup Poll. I remember the complex methods that were defined for them to go forth and gather interviews and surveys - and how impossible it was for them to actually put that into accurate practice on the ground in the field. The in-the-field sampling problems are immense. E.g. sampling in an orthodox Jewish area will give significantly different results if taken on a Thursday vs a Saturday. And whether a husband is at home and listening or at work can greatly alter responses from work-at-home women (housewives.)