Prof Gaelle Vallee-TourangeauNow I have known about base rate neglect for a few (ahem) decades. But it was only today that I heard about statistical discrimination listening to Iris Bohnet’s book on gender equality by design. Has someone made the link? Sounds to me it’s the opposite of base-rate neglect? Is that a black spot in our field? https://en.m.wikipedia.org/wiki/Statistical_discrimination_(economics) @dggoldst @UlrikeHahn @liamdelaney2020
Dan Goldstein@profgaelle @UlrikeHahn @liamdelaney2020 I didn't know about statistical discrim as the opposite of base rate neglect. Good catch. The only thing I'd heard about was 1960s research by Ward Edwards and others finding that people are conservative Bayesians, i.e., they follow base rates too much. When I searched I found a paper by Ulrike, funnily enough https://www.ucl.ac.uk/lagnado-lab/publications/harris/cornerEtAl2010.pdf
Ulrike Hahn@profgaelle @dggoldst @liamdelaney2020
that's super interesting! hadn't come across the term , I don't think.
A couple of thoughts I had over night on this:
1. using that statistical info to predict individual performance is, of course, entirely rational (and normative from a Bayesian perspective) in that it will give our 'best estimate'.
2. it's *not* normative from the basis of other values (legal and moral values).
3. the more familiar example of this tension is criminal trials.
1/
@profgaelle @dggoldst @liamdelaney2020
2/ one of the common critiques of the use of probabilities in law has been the point that one cannot use base rate information to reach a determination on the guilt of an individual, as this violates fundamental principles governing criminal trials.
4. it also reminds me of debate in the stereotype literature (on what they are).
5. it's real world evidence that 'base rate neglect' is overblown (though I've no doubt it exists...see 'conservatism lit' ;-))
@UlrikeHahn @dggoldst @liamdelaney2020 I’m glad to read it is not just me! I hadn’t thought about the legal and moral context. Very interesting indeed! For criminals, would there be less conservatism and more base rate neglect because people have strong intuitions that using base rates for convictions is wrong? 1/2
@UlrikeHahn @dggoldst @liamdelaney2020 And could these moral judgements also account for greater conservatism when the moral intuition is not so strong? (that would be why there is statistical discrimination, eg in implicit gender or racial bias inference recruitment )? 2/2