(1) True, if you test 1000 people you get 1 true positive and 10 false positives, for 11 total positives. The 10 false positives dwarf the single true positive.
(2) But, if you’re in the pool of people who test positive, you’re now 1/11 = 9.1% chance of having the disease, vs 0.1% in the general population. So the test has enriched the pool of sick people over the general population by about 100-fold, which is really good information in the real world.
This is why when you get a positive test, your doc orders another test of a different kind, to winnow out the false positives. Get a couple positives in a row from different kinds of tests, and your doc can be pretty sure you’ve got the disease.
Our present problem is people look at test like oracles, which Give The Answer immediately.
Even the oracles were confusing (https://en.wikipedia.org/wiki/Ibis_redibis_nunquam_per_bella_peribis).
]]>