Date:         Tue, 2 Feb 2010 13:32:40 -0500
Reply-To:     oloolo <dynamicpanel@YAHOO.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         oloolo <dynamicpanel@YAHOO.COM>
Subject:      Re: what statistical analysis or model I need to use when the
              median and mean are extremly different?
Comments: To: Marianne Kang <tongjunkang@HOTMAIL.COM>

your data is skewed to the upper bound
If all values of your dependent variable is strictly positive, you may try
a log transformation

without more information, it is difficult to give any advice on which stat
model is more appropriet for your data. So far, a Least Square model on the
log transformed dependent variables seems reasonable.

On Tue, 2 Feb 2010 12:44:46 -0500, Marianne Kang <tongjunkang@HOTMAIL.COM>
wrote:

>Hi all,
>Hope someone can give me some idea. I have a data with dependent variable
>is continuous but not normally distributed. The median is only around 60,
>but the mean as high as 22000,what statistical analysis or model I need to
>use when the median and mean are extremly different?
>Thanks
>Marianne