Re:My deadly disease was just a 23andme bug
by ig1 on Monday November 25, 2013 @06:20PM
(#6797803)
This is actually a stats 101 issue. Imagine that most tests only had 5% false positive rate, then if you took 20 tests then chances are at least one of those is going to give you a false positive.
[ Parent
| on HN ]
Re:My deadly disease was just a 23andme bug
by renata on Monday November 25, 2013 @06:59PM
(#6798041)
To be pedantic on stats 101, the chance of having at least one false positive is actually (1 - (.95^20)) ~= 64%, because it's 1 - the probability of every test not having a false positive.
Re:My deadly disease was just a 23andme bug
by niels_olson on Monday November 25, 2013 @08:19PM
(#6798413)
Man, that's not pedantic. That's the critical insight that demonstrates someone is competent to talk about stats at all. I wish more people would take the time to point that out when people make the OP's error.
Re:My deadly disease was just a 23andme bug
by ScottBurson on Monday November 25, 2013 @10:40PM
(#6798964)
In fairness, ig1 just said "chances are...", meaning the odds are better than 50-50.
Yes.
But no-one is going to remember "solve x for (1-(.95^x)) > 50" the purpose of the 95% & 20 is that it's easy for people to remember.