I will dig it up. Meanwhile here is an
illustration of why p-values are misleading.

Let us start with a macabre situation. What is
the probability of getting a dead person given
that the person was hanged? It will be very high
say 0.98. Now reverse the question: what is the
probability that a person has been hanged given
that he is dead? This will be quite low say
0.001. It is hardly likely you will confuse these
two probabilities and what they stand for.

Going over a simplified statistical problem in
medical research. Say a company making a new
anti-hypertensive drug has tested this drug on a
group of people and compared this group to
another similar group that has been given the
standard anti-hypertensive drug. Let the average
difference in the blood pressures between the two
groups be 10 mmHg. The statistical question that
is answered is: what is the probability of
obtaining an average difference of 10 mmHg
between the two groups given that the null
hypothesis is true? The null hypothesis is that
the effect of the two drugs on blood pressure is
the same. Let us say that this probability is
0.04. This is the so called p-value and since
this value is below 0.05 you declare that “it is
significant”. Now reverse the question: what is
the probability that the null hypothesis is true
given the average difference is 10 mmHg between
the two groups? This is the question you (or the
study investigator) are presumably interested in
answering – this is not the p-value. If you take
this to be the p-value then you are confusing the
two probabilities depicted in the previous
paragraph. 

In a nutshell, the p-value gives the correct
answer to the wrong question. 

- Shyam

--- shiv sastry <[EMAIL PROTECTED]> wrote:

> 
> Shyam could you post/link that very interesting
> paper you had put on 
> jipmer.net about the guy who came up with this
> p value concept?



      
____________________________________________________________________________________
Don't let your dream ride pass you by. Make it a reality with Yahoo! Autos.
http://autos.yahoo.com/index.html
 




Reply via email to