Date: Fri, 20 Jun 2008 17:25:14 -0400
Reply-To: "Amy T. Brown" <amara_tabu@YAHOO.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: "Amy T. Brown" <amara_tabu@YAHOO.COM>
Subject: Units on independent variables
I am doing a ZIP regression. I have 4 predictor variables, A, B, C, and D.
A and B are ratios. C and D are $ amounts in millions. Since it is
executive ownership data, I have some huge values for C and D. And later
on I include a variable E that is the square of D. And then ZIP has
problems with calculating covariance and correlation. My question is that
if I divide C and D by 1000, to essentially report them as in billions of
dollars, would that affect my regression results? Because, I mean, they
could be in billions of dollars to start with, right?
Thank you so much!