Date: Fri, 2 Apr 2010 15:51:18 -0400
Reply-To: oloolo <dynamicpanel@YAHOO.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: oloolo <dynamicpanel@YAHOO.COM>
Subject: Re: SAS Probit Xbeta linear prediction
since you used CLASS statement for your dependent variable "p", make sure
the design in SAS is the same as in STATA, ie. they have the same reference
class, otherwise their output of Xbeta will be different.
On Thu, 1 Apr 2010 12:40:15 -0400, McSutlt Tim <sutltlast@YAHOO.COM> wrote:
>Hi guys, could anyone help me to solve the following problem.
>I use the proc probit in sas and try to get the linear prediction with the
>PROC PROBIT ORDER=DATA;
>MODEL p=a b c;
>OUTPUT OUT=mypredict XBETA=pred_sas;
>I use option XBETA to get the linear predictions
>However, I find the predictions derived from SAS are quite different from
>those derived from STATA.
>PROBIT p a b c;
>predict pred_stata, xb;
>The coefficients on independent var. are the same, but pred_sas are quite
>different from pred_stata.
>Calculating manually, I felt that pred_stata are really the linear
>prediction I want.
>Could someone tell me why they're different?
>Many thanks in advance!