|
Shroomerited


Registered: 06/12/10
Posts: 1,974
Last seen: 7 years, 10 months
|
Bayes' Theorem applied to drug testing
#18745103 - 08/23/13 12:19 AM (10 years, 6 months ago) |
|
|
http://en.wikipedia.org/wiki/Bayes'_theorem#Drug_testing
If you ever fail a test, just tell them you're more likely to NOT be a drug user than to be a drug user.
|
ModestMouse
IM WALKIN ON SUNSHINE


Registered: 05/06/13
Posts: 19,227
Loc: Upstate
|
Re: Bayes' Theorem applied to drug testing [Re: Shroomerited]
#18745150 - 08/23/13 12:30 AM (10 years, 6 months ago) |
|
|
-------------------- Anyone got a lowpass filter in this biiiiash?
|
Led Zeppelin
Tripper


Registered: 05/17/10
Posts: 3,962
Last seen: 3 years, 5 months
|
Re: Bayes' Theorem applied to drug testing [Re: ModestMouse]
#18745295 - 08/23/13 01:14 AM (10 years, 6 months ago) |
|
|
the only thing I understood was the part about speaking to the person on the train. common sense?
|
Shroomerited


Registered: 06/12/10
Posts: 1,974
Last seen: 7 years, 10 months
|
Re: Bayes' Theorem applied to drug testing [Re: Led Zeppelin]
#18745306 - 08/23/13 01:18 AM (10 years, 6 months ago) |
|
|
Just read the part about the drug test. If a test is "99% accurate" and .5% of the population uses a drug and you test positive for it, you have a a 33% chance of actually being a drug user. So you're more likely to not be a drug user.
|
Shroomerited


Registered: 06/12/10
Posts: 1,974
Last seen: 7 years, 10 months
|
Re: Bayes' Theorem applied to drug testing [Re: Shroomerited]
#18745311 - 08/23/13 01:21 AM (10 years, 6 months ago) |
|
|
This is only if .5% of the population uses the drug. If 10% uses the drug, there's a 93% chance you used it.
|
Konyap

Registered: 06/30/07
Posts: 33,945
Loc: Planet Piss
Last seen: 4 years, 3 months
|
Re: Bayes' Theorem applied to drug testing [Re: Shroomerited]
#18745341 - 08/23/13 01:35 AM (10 years, 6 months ago) |
|
|
I fucking hated probability I got an average of like 40 in it at some point.
|
dontknow
It's all in the reflex


Registered: 07/05/13
Posts: 3,889
Last seen: 7 years, 4 months
|
Re: Bayes' Theorem applied to drug testing [Re: Shroomerited]
#18745342 - 08/23/13 01:36 AM (10 years, 6 months ago) |
|
|
i can't rap my head around this one
--------------------
The discerning heart seeks knowledge, but the mouth of a fool feeds on folly. -Proverbs 15:14
“Imagination is everything. It is the preview of Life’s coming attractions.” Albert Einstein
|
Konyap

Registered: 06/30/07
Posts: 33,945
Loc: Planet Piss
Last seen: 4 years, 3 months
|
Re: Bayes' Theorem applied to drug testing [Re: Konyap]
#18745345 - 08/23/13 01:38 AM (10 years, 6 months ago) |
|
|
I understand the formula it's just this is a very odd scenario.
|
Shroomerited


Registered: 06/12/10
Posts: 1,974
Last seen: 7 years, 10 months
|
Re: Bayes' Theorem applied to drug testing [Re: Konyap]
#18745348 - 08/23/13 01:39 AM (10 years, 6 months ago) |
|
|
Quote:
aiyobro said: I understand the formula it's just this is a very odd scenario.
Mhmm. This.
|
Konyap

Registered: 06/30/07
Posts: 33,945
Loc: Planet Piss
Last seen: 4 years, 3 months
|
Re: Bayes' Theorem applied to drug testing [Re: Konyap]
#18745350 - 08/23/13 01:39 AM (10 years, 6 months ago) |
|
|
This surprising result arises because the number of non-users is very large compared to the number of users, such that the number of false positives (0.995%) outweighs the number of true positives (0.495%). To use concrete numbers, if 1000 individuals are tested, there are expected to be 995 non-users and 5 users. From the 995 non-users, 0.01 × 995 ≃ 10 false positives are expected. From the 5 users, 0.99 × 5 ≃ 5 true positives are expected. Out of 15 positive results, only 5, about 33%, are genuine.
|
Konyap

Registered: 06/30/07
Posts: 33,945
Loc: Planet Piss
Last seen: 4 years, 3 months
|
Re: Bayes' Theorem applied to drug testing [Re: Konyap] 2
#18745364 - 08/23/13 01:48 AM (10 years, 6 months ago) |
|
|
Wow this makes that mandatory welfare drug testing look bad.
|
Led Zeppelin
Tripper


Registered: 05/17/10
Posts: 3,962
Last seen: 3 years, 5 months
|
Re: Bayes' Theorem applied to drug testing [Re: Konyap]
#18745510 - 08/23/13 03:10 AM (10 years, 6 months ago) |
|
|
I mean, is this really practical though? "yes, I know boss, I failed a test for cocaine, but you SEE, the probabilty of the percentage of the blah blah blah"
|
druqs
ALKALOIDOHOLIC


Registered: 09/11/06
Posts: 8,862
Last seen: 5 months, 12 hours
|
Re: Bayes' Theorem applied to drug testing [Re: Shroomerited]
#18745519 - 08/23/13 03:17 AM (10 years, 6 months ago) |
|
|
sounds like op is grasping at straws.
|
Shroomerited


Registered: 06/12/10
Posts: 1,974
Last seen: 7 years, 10 months
|
Re: Bayes' Theorem applied to drug testing [Re: Led Zeppelin]
#18746784 - 08/23/13 11:08 AM (10 years, 5 months ago) |
|
|
Quote:
Led Zeppelin said: I mean, is this really practical though? "yes, I know boss, I failed a test for cocaine, but you SEE, the probabilty of the percentage of the blah blah blah"
Of course not. If 10% of the population uses the drug and you fail a drug test, you have a 93% chance of being a drug user. Drug tests work because there's so many drug users.
|
Sophistic Radiance
Free sVs!



Registered: 07/11/06
Posts: 43,135
Loc: Center of the Universe
|
Re: Bayes' Theorem applied to drug testing [Re: Shroomerited]
#18746788 - 08/23/13 11:09 AM (10 years, 5 months ago) |
|
|
Quote:
Shroomerited said: Just read the part about the drug test. If a test is "99% accurate" and .5% of the population uses a drug and you test positive for it, you have a a 33% chance of actually being a drug user. So you're more likely to not be a drug user.
-------------------- Enlil said: You really are the worst kind of person.
|
badchad
Mad Scientist

Registered: 03/02/05
Posts: 13,373
|
|
I don't think that will work. In the case of drug testing, the worst scenario is a "false positive". This is akin to accusing someone of using, when they don't. The alternative, a "false negative", has lesser consequence.
It's similar to the philosophy of you'd rather let a guilty man go free, than send an innocent man to jail. We're much more concerned about the false positive in this case.
In this hypothetical scenario, the chances of a false positive are 1%.
It doesn't matter how many people use or don't use, the chance are 1%, period.
-------------------- ...the whole experience is (and is as) a profound piece of knowledge. It is an indellible experience; it is forever known. I have known myself in a way I doubt I would have ever occurred except as it did. Smith, P. Bull. Menninger Clinic (1959) 23:20-27; p. 27. ...most subjects find the experience valuable, some find it frightening, and many say that is it uniquely lovely. Osmond, H. Annals, NY Acad Science (1957) 66:418-434; p.436
|
volcomstoner
I'll have just one more xanax



Registered: 07/20/09
Posts: 11,231
Loc: Minnesnowta
|
Re: Bayes' Theorem applied to drug testing [Re: badchad]
#18746972 - 08/23/13 11:53 AM (10 years, 5 months ago) |
|
|
I had a false negative once, I was pissed though because I needed to fail to get into treatment.
--------------------
HAIL SATAN Vas donc jouer dans le traffic
|
badchad
Mad Scientist

Registered: 03/02/05
Posts: 13,373
|
Re: Bayes' Theorem applied to drug testing [Re: volcomstoner]
#18746983 - 08/23/13 11:56 AM (10 years, 5 months ago) |
|
|
Quote:
volcomstoner said: I had a false negative once, I was pissed though because I needed to fail to get into treatment.
In clinical diagnostic tests, often times failure rates are designed around the worst possible scenario. For example, one might argue that for HIV, a "false negative" is the worst outcome.
-------------------- ...the whole experience is (and is as) a profound piece of knowledge. It is an indellible experience; it is forever known. I have known myself in a way I doubt I would have ever occurred except as it did. Smith, P. Bull. Menninger Clinic (1959) 23:20-27; p. 27. ...most subjects find the experience valuable, some find it frightening, and many say that is it uniquely lovely. Osmond, H. Annals, NY Acad Science (1957) 66:418-434; p.436
|
|