Numbers to think about
 

Let's assume that the labs and their tests are 99% accurate.

The UCI did around 12000 tests last year, and about 380 came back
positive. These are just rough numbers off the top of my head.
It worked out to around 3.8% of all tests came back positive.

So, if you take that 99% accuracy number and apply it,
you end up with roughly 1 out of 3 positives due to bad testing.

Numbers to think about
 

"CowPunk" a écrit dans le message de news:
...
Let's assume that the labs and their tests are 99% accurate.

The UCI did around 12000 tests last year, and about 380 came back
positive. These are just rough numbers off the top of my head.
It worked out to around 3.8% of all tests came back positive.

So, if you take that 99% accuracy number and apply it,
you end up with roughly 1 out of 3 positives due to bad testing.


???

Tell it again ....

It is not the way I learned math ; )

If the test are 99 % accurate (positive or negative) so 1 % (positive or
negative) are not.

If 380 came back positive and 1 % are not accurate, it is to say that 3.8 ,
let say 4 are not.

1.05 % are not accurate, it is to say 4 out of 380

Where you 1 of 3 comes from ?????

Numbers to think about
 

CowPunk wrote:
Let's assume that the labs and their tests are 99% accurate.

The UCI did around 12000 tests last year, and about 380 came back
positive. These are just rough numbers off the top of my head.
It worked out to around 3.8% of all tests came back positive.

So, if you take that 99% accuracy number and apply it,
you end up with roughly 1 out of 3 positives due to bad testing.


It's been a million years since I took a probability class so I must
have just confused myself. Someone please straighten me out here.

If the probability of a false positive is .01 then the probability of
both A and B samples receiving a false positives is .01 * .01 = .0001.
I think that means that ~1.2 times a year someone innocent should fail
both the A and B sample despite being clean.

That's got to be wrong.

Numbers to think about
 

1% of 12000 = 120

120:380 ~ 1:3

Numbers to think about
 

"Tim Lines" a écrit dans le message de news:
...
CowPunk wrote:
Let's assume that the labs and their tests are 99% accurate.

The UCI did around 12000 tests last year, and about 380 came back
positive. These are just rough numbers off the top of my head.
It worked out to around 3.8% of all tests came back positive.

So, if you take that 99% accuracy number and apply it,
you end up with roughly 1 out of 3 positives due to bad testing.


It's been a million years since I took a probability class so I must have
just confused myself.

I too

Someone please straighten me out here.

If the probability of a false positive is .01 then the probability of both
A and B samples receiving a false positives is .01 * .01 = .0001. I think
that means that ~1.2 times a year someone innocent should fail both the A
and B sample despite being clean.

That's got to be wrong.

Let put it in other way.



We have 12000 test and 1% have a wrong result. 1% out of 12000 = 120.



In the 120 we have some Good Guys wrongly called cheaters, and some cheaters
called Good Guys

OK ?



Let see the distribution of this mistake :



380 Positive * 1% = 3.8 ( so 3.8 out of 380 are clean guys called cheaters)

12000-380= 11620 Negative * 1 % = 116.2 (so 116.2 are cheaters but found
Good guys.)



Let see if there is a mistake . 116.2 + 3.8 = 120



Ok 120 is what we expected.



In short around 4 out 380 or 1 out 95 are poor guys called cheaters but they
are not..

On other side 116 out of 11620 or around 1 out of 100 are lucky cheaters



Once again why did you said 1 out of 3 ????

Numbers to think about
 

If the probability of a false positive is .01 then the probability of both
A and B samples receiving a false positives is .01 * .01 = .0001. I think

No I said 99% accuracy. Errors could be based on mishandling sample,
contamination, etc.... I just don't believe that a lab is 99.9%
accurate in their work.


We have 12000 test and 1% have a wrong result. 1% out of 12000 = 120.
Yes


380 Positive * 1% = 3.8 ( so 3.8 out of 380 are clean guys called cheaters)

So now you are applying 1% again.
Which means you are calculating based 0.1% accuracy. 1%x1%


Where we are diverging is you are applying 1% to the positives, while I
am applying 1%
to the total # of tests. IMHO, Accuracy of a test applies to the total
# of tests performed.

Numbers to think about
 

"CowPunk" a écrit dans le message de news:
...

If the probability of a false positive is .01 then the probability of
both
A and B samples receiving a false positives is .01 * .01 = .0001. I
think

No I said 99% accuracy. Errors could be based on mishandling sample,
contamination, etc.... I just don't believe that a lab is 99.9%
accurate in their work.


We have 12000 test and 1% have a wrong result. 1% out of 12000 = 120.
Yes


380 Positive * 1% = 3.8 ( so 3.8 out of 380 are clean guys called
cheaters)

So now you are applying 1% again.
Which means you are calculating based 0.1% accuracy. 1%x1%

Of course no. The figure 380 of positive result is your, not the result of
some 1%


Where we are diverging is you are applying 1% to the positives, while I
am applying 1%
to the total # of tests. IMHO, Accuracy of a test applies to the total
# of tests performed.

Numbers to think about
 

CowPunk wrote:
Let's assume that the labs and their tests are 99% accurate.

The UCI did around 12000 tests last year, and about 380 came back
positive. These are just rough numbers off the top of my head.
It worked out to around 3.8% of all tests came back positive.

So, if you take that 99% accuracy number and apply it,
you end up with roughly 1 out of 3 positives due to bad testing.

I'm not a medical technician, and I don't play one on TV, but I have
heard from reliable sources that the false-positive and false-negative
rates in medical testing can be substantially different. For all I
know, this might be the rule rather than the exception.

An illustration with made-up numbers: Some test might have a false
positive rate of 10% (10% of those who are really "negative" are deemed
"positive" by the test) while only returning a 3% false negative rate
(only 3% of thoses truly "positive" are "missed" by the test). Again,
these numbers are entirely made up, only to illustrate the phenomenon.

Mark

Numbers to think about
 

Montesquiou wrote in
:


"Tim Lines" a écrit dans le message de news:
...
CowPunk wrote:
Let's assume that the labs and their tests are 99% accurate.

The UCI did around 12000 tests last year, and about 380 came
back positive. These are just rough numbers off the top of
my head. It worked out to around 3.8% of all tests came back
positive.

So, if you take that 99% accuracy number and apply it,
you end up with roughly 1 out of 3 positives due to bad
testing.


It's been a million years since I took a probability class so
I must have just confused myself.

I too

Someone please straighten me out here.

If the probability of a false positive is .01 then the
probability of both A and B samples receiving a false
positives is .01 * .01 = .0001. I think that means that ~1.2
times a year someone innocent should fail both the A and B
sample despite being clean.

That's got to be wrong.

Let put it in other way.



We have 12000 test and 1% have a wrong result. 1% out of 12000 =
120.



In the 120 we have some Good Guys wrongly called cheaters, and
some cheaters called Good Guys

OK ?

Let see the distribution of this mistake :

380 Positive * 1% = 3.8 ( so 3.8 out of 380 are clean guys
called cheaters)

12000-380= 11620 Negative * 1 % = 116.2 (so 116.2 are cheaters
but found Good guys.)

Let see if there is a mistake . 116.2 + 3.8 = 120

Ok 120 is what we expected.

In short around 4 out 380 or 1 out 95 are poor guys called
cheaters but they are not..

On other side 116 out of 11620 or around 1 out of 100 are lucky
cheaters

Once again why did you said 1 out of 3 ????


Please see:

http://yudkowsky.net/bayes/bayes.html

Numbers to think about
 

"CowPunk" a écrit dans le message de news:
...
1% of 12000 = 120

120:380 ~ 1:3


Oh my friend !!!

With all due respect if it is way they teach statistic in your country ...
You are lost.

However as I have many friends in the USA and I know they are not so
ignorants in Math, I believe the problem is your.

Since your original post you DECIDED that 1% of the test were wrong.

So 1% of the 380 positive (that you DECIDED BY YOUR OWN) are wrong.

1% of 380 is 3.8.

Turn your problem the way you want 1% is allway 1% and NEVER 1:3 (33.33 %)
!!

Oh my God, pls help me !