[FoRK] Black Belt Bayesian vs. Authority, Fight!
<baisley at alumni.rice.edu> on
Sun Aug 12 22:25:10 PDT 2007
> > The correct answer is 7.8%.
> > Hmm. I get 8.4%. How is this problem worked out?
[What I meant to say is ...]
The distribution of those who test positive or negative versus those who
don't/have it is:
> Neg Pos
> No 89.5% 9.5%
> Yes 0.2% 0.8%
> 7.8% of the 10.3% who test positive will actually have cancer. Still,
> it's nicer to test negative.
And, as has been noted before, doctors are generally lousy at statistics.
More information about the FoRK