Donate SIGN UP

More Probabilities ...

Avatar Image
sunny-dave | 21:21 Wed 29th Oct 2014 | ChatterBank
68 Answers
... but very practical ones this time.


You are invited to go for a screening test for an illness - you have no symptoms.

About 1% of the screened group will have the illness.

The test is 99% accurate at diagnosing people who have the illness.

You get a 'positive' result.

How likely is it that you actually have the illness?


99%?

That's what many people think.


Gravatar

Answers

41 to 60 of 68rss feed

First Previous 1 2 3 4 Next Last

Best Answer

No best answer has yet been selected by sunny-dave. Once a best answer has been selected, it will be shown here.

For more on marking an answer as the "Best Answer", please visit our FAQ.
Shoota, you've gone pink and blobby!
That's because 33% of me isn't feeling myself, (or was it 66%?)
If a vacuum has to be filled, Woof....what would fill the one left by AnswerBank......if the ant was stepped on?

'Sallright, Psyb.....pink and blobby but still active....x
But 'limp' as well?
I like coffee in mine.
Or zoop.
Has Dave abandoned us?
Probably.
shoota has a girlfriend :(
NOW HE HAS A BABY ???
Boy or girl?
Was there a reason to ask the question and then give the answer before folk had a chance to think ?

I'd advise them to tell everyone they had the illness. They'd get a 100% accurate result at diagnosing ill folk then. And you can't beat that.
Oh Dear Sonny
I could have warned you against this
the predictable flood of "whaaaaa?" 'answers' wind out like a motorway to oblivion in front of you.

actually from your post it is reasonable assume that the false neg rate is zero cos you didnt mention it and on your figures then

of a 1 000 tested, 100 have the disease, another one percent dont have the disease but test positive which is 90,
so of the 190 who have tested positive, 100 have it so the PPV is around one half ( if you have a positive test the chance of having the disease is around 1/2 )

so yup your point is that in a screening test
you need to know the PPV - positive predictive value

pt well made and not understood by the huddled masses who should be out at work or feeding the cats

why not get the cat fur flying by asking if breast screening works?
PPV of mammography is pretty crap in fact

or get into PSA tests - these return a value on a continuum, and there is a cut off value. Obviously set it too low and you will flood the surgeries with worried well, and set it too high and men with the disease will croak when you reassure them they are cancer free.
You need rocks to get over that one ( ROCs - receiver operating curves, and that is really taxing )

honestly if you asked, I would have told you not to do this !
thanks pp. it makes complete sense now !!!
'Oh Dear Sonny
I could have warned you against this
the predictable flood of "whaaaaa?" 'answers' wind out like a motorway to oblivion in front of you.

actually from your post it is reasonable assume that the false neg rate is zero cos you didnt mention it and on your figures then

of a 1 000 tested, 100 have the disease, another one percent dont have the disease but test positive which is 90,
so of the 190 who have tested positive, 100 have it so the PPV is around one half ( if you have a positive test the chance of having the disease is around 1/2 )

so yup your point is that in a screening test
you need to know the PPV - positive predictive value

pt well made and not understood by the huddled masses who should be out at work or feeding the cats

why not get the cat fur flying by asking if breast screening works?
PPV of mammography is pretty crap in fact

or get into PSA tests - these return a value on a continuum, and there is a cut off value. Obviously set it too low and you will flood the surgeries with worried well, and set it too high and men with the disease will croak when you reassure them they are cancer free.
You need rocks to get over that one ( ROCs - receiver operating curves, and that is really taxing )

honestly if you asked, I would have told you not to do this !'


Pedant!






Probably.....
I think your probably right, shoota !
hang on hang on
The statement made by SD is that the test is "99% accurate at diagnosing the illness" surely that statement, if true, must take into account the rate of false positives and allow for them otherwise the screeners are just plain lying?
Because doesn't that "99% accuracy statement mean" "in 99 cases out of a hundred, when the test comes back positive, the result is accurate and the person tested does have the illness."?
The test is 99% accurate at diagnosing people who have the illness.
i'm still trying to decide which sack to pick ...
you are right OG, have just re read and that is more ambivalent.

41 to 60 of 68rss feed

First Previous 1 2 3 4 Next Last

Do you know the answer?

More Probabilities ...

Answer Question >>