I like this approach using tabi as a calculator for this.
r(p_exact) is P(n11 >= 3), the p-value in the Fisher's exact test
if we calculate P(n11 >= 4), we can obtain P(n11 = 3) by difference
. tabi 4 0 \ 3 113, exact
           |          col
       row |         1          2 |     Total
-----------+----------------------+----------
         1 |         4          0 |         4 
         2 |         3        113 |       116 
-----------+----------------------+----------
     Total |         7        113 |       120 
           Fisher's exact =                 0.000
   1-sided Fisher's exact =                 0.000
. return list
scalars:
           r(p1_exact) =  4.26072210718e-06
            r(p_exact) =  4.26072210718e-06
                  r(c) =  2
                  r(r) =  2
                  r(N) =  120
     |  p3exact     p3plus     p4plus |
     |--------------------------------|
  1. | .0004815   .0004857   4.26e-06 |
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Visintainer PhD,
Paul
Sent: Wednesday, October 28, 2009 7:49 AM
To: '[email protected]'
Subject: st: RE: probability question
I'm probably thinking of this too simplistically, but wouldn't this just be
a Fisher's exact test?
. tabi 3 1 \ 4 112, exact   /* where col = rain and row = hat */
           |          col
       row |         1          2 |     Total
-----------+----------------------+----------
         1 |         3          1 |         4 
         2 |         4        112 |       116 
-----------+----------------------+----------
     Total |         7        113 |       120 
           Fisher's exact =                 0.000
   1-sided Fisher's exact =                 0.000
. return list
scalars:
           r(p1_exact) =  .0004857223202188
            r(p_exact) =  .0004857223202188
                  r(c) =  2
                  r(r) =  2
                  r(N) =  120
___________________________________
Paul F. Visintainer, PhD
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Richard Goldstein
Sent: Wednesday, October 28, 2009 8:38 AM
To: statalist
Subject: st: probability question
it's been a long time since I thought about questions like this, but, as
a lead-in to a study, a client has asked the following question which he
thinks he understands and says is related to where he wants to go:
during a consecutive period of 120 days, if it rains on 7 days and my
client wears a hat on 4 days (these are independent of any knowledge of
the weather), what is the probability that it will rain on 3 of the days
on which he is wearing a hat?
my client swears that this is not a homework problem for him or his wife
or one of their kids!
Rich