| Date: | Mon, 13 Nov 2000 19:51:17 -0500 |
|---|
| Reply-To: | Simcha Pollack <pollacks@STJOHNS.EDU> |
|---|
| Sender: | "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU> |
|---|
| From: | Simcha Pollack <pollacks@STJOHNS.EDU> |
|---|
| Subject: | Presidential elections: Bug in Binomial? |
|---|
| Content-Type: | text/plain; charset="us-ascii" |
|---|
I am using SAS v6.12 t o figure the probability of getting a result as
close as was obtained in Florida and NM.
I compute the cumualtive binomial as:
x=probbnml(.5 , 6000000 , 299650);
which comes out to approxiamtely .40 .
Thus, the probability of the election in Florida being within 350 (either
way) is .2. This seems unreasonably high. Is it correct?
Secondly, when I use the probbnml function with 300100 as the third
argument it goes down! Is this a known bug in SAS?
Simcha Pollack
| 
