> Isn't there some statistics theorem that
says there's always way more false positives than
false negatives?
Suppose there's a condition
that affects 1% of the population, and also that
there's a diagnostic test that's 99% effective and
assigns false positives and false negatives with equal
probability.
Then:
P[false positive] = P[negative]*P[error]
P[false
positive] = .99*.01
P[false positive] =
.0099
P[false negative] = P[positive]*P[error]
P[false
negative] = .01*.01
P[false negative] = .0001
So
a false positive is 99 times more likely than a
false negative. This occurs because the relative
incidence of the condition is small and that we expect more
errors in the larger, negative group than in the
smaller, positive one because, well, it's
larger.
Dave
(This test, most likely, doesn't have 99% accuracy)