> It puts you in the company of a lot of highly educated doctors.
Too bad for them, because it's really the most basic stuff about statistical inference, something that every professor tells on the first lecture, classic example, something like Monty Hall problem or birthday paradox. Not that it is surprising, because very few people know something properly at all, be they PhD or butchers… But no more comforting, really.
Of 10 medical students given the quiz, only two got the right answer. So we can hope that the other eight will flunk medical school and never treat any patients.
Screw this ethos - "One failure? I hope you are fired". Heaven forfend the idea of better education or refresher classes/requirements.
The simple fact is that mathematical problem solving is a rarely used skill for doctors and most professions which require extensive schooling. I'm confident the doctor in this case would immediately test multiple times to rule out false positives, just because thats what the medical literature on this rare disease says to do. And if he didn't, it would be a mistake, which happens a lot in medicine, but it doesn't have much to do with doctors ability to solve math problems.
Would that do any good though? Are failures correlated in the case of the Alzheimer's test? What causes the test to fail? If it's truly random, then yes, you can retry; but that presupposes that trials are independent of each other; and I'm be surprised if that were in fact the case.
I can't understand why I'm not getting 2% from my point of view.
A person is identified positive in two distinct cases: if it is ill (p1=0.001) AND the test shows positive (q1=0.95), or when it is clean (p2=0.999) but the test wrongly shows positive (q2=0.05). The probability of this event is P=p1 * q1+p2 * q2=0.001 * 0.95+0.999 * 0.05=0.0509
8 comments
[ 3.5 ms ] story [ 32.4 ms ] threadReminds me of this classic Zed Shaw rant.
http://zedshaw.com/essays/programmer_stats.html
Too bad for them, because it's really the most basic stuff about statistical inference, something that every professor tells on the first lecture, classic example, something like Monty Hall problem or birthday paradox. Not that it is surprising, because very few people know something properly at all, be they PhD or butchers… But no more comforting, really.
Screw this ethos - "One failure? I hope you are fired". Heaven forfend the idea of better education or refresher classes/requirements.
Would that do any good though? Are failures correlated in the case of the Alzheimer's test? What causes the test to fail? If it's truly random, then yes, you can retry; but that presupposes that trials are independent of each other; and I'm be surprised if that were in fact the case.