|Date: ||Tue, 25 Oct 2005 23:28:19 -0700|
|Reply-To: ||David L Cassell <davidlcassell@MSN.COM>|
|Sender: ||"SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>|
|From: ||David L Cassell <davidlcassell@MSN.COM>|
|Subject: ||Re: Skewed variables & surveys|
|Content-Type: ||text/plain; format=flowed|
>I don't understand why some of the answers to this question imply that OLS
>is invalid if the dependent variable is skewed or not continuous. OLS is
>based on the distribution of errors for the correct linear model, so
>everything is relative to the independent variables. Can't say much until
>know about these variables.
>Also-- just because most of the survey answers are 5 6 or 7 does not make
>the variable skewed-- could still be symmetric around 6 or even 5.5
I see Peter has already jumped all over oyu, so let me point out a couple
 You're right that it is the form of the errors that matters, not the
of the dependent variable. BUT when the dependent variable is a categorical
variable with a small number of values, and the independent variables are
restricted in values, then the behavior of the error terms is fairly
 When the survey values go from 1 to 7, and 75% of the values are placed
on 6 and 7, then that *does* mathematically force the data to be skewed.
Plus the large point masses on '6' and '7' will tend to force the model into
fairly specific forms for the error terms.
It used to be that *everyone* did OLS with data like this. But that was
they had to given the state of computers and computer programs, not because
they ought to do it. Even after code became available, people did *not*
change, and sub-optimal recommendations are still being made to trusting
That's why people like me get to be such enormous pains in the, umm, eye.
David L. Cassell
3115 NW Norwood Pl.
Corvallis OR 97330
Is your PC infected? Get a FREE online computer virus scan from McAfeeŽ