Replies to post #128717 on Biotech Values

[exwannabe]

Re: Clark problem

Obviously your math is correct.

I thought that Clark was going for the false negatives when he first posted, but was surprised when he tossed out 30% rate, as 1/3 is the cut point at which the numbers flop.

So either:

Clark is using 30% to get the numbers close, and is saying there is still some minor unexplained effect we are both missing.

Or

Clark mistyped 30% and really meant 35% (which flips your math to B). In that case there still is another main point that I missed on the first cut that Clark still is looking for. If so I have a good answer (which would of course flip it back to A), but will keep my mouth shut.

[iwfal]

Ok - that is the math that you'd expect because it is the answer that almost everyone gives. But it isn't actually correct. Hopefully the below question will illustrate why. But before calculating the correct answer as per the below question let me re-emphasize that your original (incorrect) answer had Drug A and Drug B as almost exactly equal in risk.

Remaining question - So... what if I told you that the true incidence of the disease among those tested is 1% (e.g. if you tested 10,000 people but treated no one then 100 would die of the disease). And the False Negative rate is 10%.


PS Bonus question - what is the error that you made in your math (an error that almost everyone makes)?