Prev: Nomograms using annotate?
Next: rollup
From: Gary Klein on 18 Jan 2010 15:05
Your estimates (and standard errors) are for the log(odds), that is,

estimate = log(odds) = log(p/(1-p)).

So, you just need to solve for p to give you the probability:

p = exp(estimate) / (1 + exp(estimate)).

Hope this helps.





--- On Mon, 1/18/10, Val Krem <valkrem(a)YAHOO.COM> wrote:

> From: Val Krem <valkrem(a)YAHOO.COM>
> Subject: result
> To: SAS-L(a)LISTSERV.UGA.EDU
> Date: Monday, January 18, 2010, 11:45 AM
> Hi all,
>
>
>
>
> � I have the following data set :-
>
>
> Color has two classes (white or black) white=1 ;black=0;
>
> Texture soft or hard (1 soft and 0=hard); * is response
> variable.
>
> I fitted the follwing model,
> Proc logistic data=xx;
>
> Model texture (event=”1”)=color;
>
> Ods output paramterEstimates=model_color;
>
> �
>
> � � � � � � � � � � � � � � � � � � � � � Estimate
> � � std.error
>
> � (Intercept) � � 12.052� � � 3.11037 � �
>
> color� � � � � � � � � � � � � 0.26180� � �
> 0.06624
>
> How do I get the probability of texture ( being � soft) for
> the two different colors(white and
> black)?
>
> � Thanks in advance,
>
>
>
>
>
 | 
Pages: 1
Prev: Nomograms using annotate?
Next: rollup