View Single Post
Old 05-31-2006, 08:38 AM  
one eight
Confirmed User
 
Join Date: Mar 2005
Location: Costa Rica & San Francisco
Posts: 1,941
I think you also should consider the very small sample data set your using. In the mainstream stuff I was doing, we'd send hundreds of thousands of hits before even determining a ratio. Less sample data = greater fluctuations = less real understanding.
__________________
-more than a number
one eight is offline   Share thread on Digg Share thread on Twitter Share thread on Reddit Share thread on Facebook Reply With Quote