View Single Post
  #20 (permalink)  
Old 03-July-2007, 08:30 PM
aurora's Avatar
aurora aurora is online now
Established Member
 
Join Date: Sep 2003
Posts: 3,064
Default

Quote:
Originally Posted by JohnW View Post
No it doesn't. You don't have binomial data. Rainfall amounts for individual days may well be exponentially distributed (I'm guessing - I'm not a meteorologist), in which case the sums and/or averages will have a gamma distribution.
Yes, in order for data to be binomial, it has to have only two mutually exclusive outcomes for each trial. Like heads or tails, true or false. Not a rainfall measurement.

And the outcomes have to be independent, daily rainfall is not, since weather systems can last for more than one day.

And the process has to be stationary, with the probability of success constant from trial to trial. So, no shifts in climate.

Strike three.
__________________
"I'm as accurate as any psychic. And I'm a cartoon!" -- Squidward

"Arrrgh, the laws of physics be a harsh mistress!" -- Bender