From: usenet2@vyznev.invalid   
      
   On 2009-05-11, James A  Donald  wrote:   
   > Ilmari Karonen   
   >> Not necessarily, since it would be quite possible for a group to be   
   >> _both_ worse at a given job _and_ also irrationally discriminated   
   >> against.   
   >   
   > But it is not possible for those hired despite being irrationally   
   > discriminated against to be worse at the given job - which is the   
   > situation I observe for female engineers.   
      
   Actually, it is.  Let me illustrate with a simplified example:   
      
   Assume that the ability to do job X can be measured objectively on a   
   scale from 0 to 100.  Further assume that applicants for the job fall   
   into two groups A and B, and that the scores for group A are uniformly   
   distributed between 0 and 70 and the scores for group B between 30 and   
   100.  We'll further assume that an employee will have complete and   
   accurate information on each applicant's ability score, as well   
   whether they belong to group A or B.  (These assumptions are hardly   
   realistic, but they simplify the math to a level one can reasonably   
   present in a Usenet post.)   
      
   Given full information of the abilities of each applicant, the   
   rational way to select employees is to choose exactly those applicants   
   whose ability score exceeds some threshold (to be set according to the   
   supply and demand of workers).  Let us assume, however, that a   
   particular employer is irrationally discriminating against group A,   
   such that they'll hire members of group A if their ability score   
   exceeds 60 but will accept members of group B even if their ability   
   score is only 40.   
      
   Then the mean ability score of hired employees in group A will be   
   (60+70)/2 = 65, while the mean ability score for employees in group B   
   will be (40+100)/2 = 70.  Thus, even though the discrimination against   
   group A is irrational (given the availability of full information on   
   actual ability), employees in group A hired despite the discrimination   
   still have lower average ability scores than those in group B.   
      
      
   >> For all we know, it *might* be that the discrimination is almost   
   >> nonexistent, but their test resumé just happened to fall on the   
   >> boundary of hireability where it made a difference.   
   >   
   > Except that they did not directly test whether it made any difference   
   > at all To actually measure the difference, they needed to report the   
   > difference in interview requests for the fictitious candidates.   
   > Instead they interviewed the people reviewing the resumes, and   
   > concluded that those people were extraordinarily and improbably   
   > prejudiced - which may well reflect the biases of the researchers,   
   > rather than the biases of those the researchers interviewed.   
      
   It would be better if you criticized the way in which the study was   
   actually conducted, rather than the entirely different study you seem   
   to have made up inside your head.  That way someone might actually   
   consider your criticism relevant.   
      
   Actually reading the paper would probably help.   
      
   --   
   Ilmari Karonen   
   To reply by e-mail, please replace ".invalid" with ".net" in address.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)