Date:         Wed, 8 Dec 1999 08:39:26 EST
Reply-To:     Jesse Chittams <jesse@LANDRU.CPR.UPENN.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Jesse Chittams <jesse@LANDRU.CPR.UPENN.EDU>
Organization: Center for Psychotherapy Research
Subject:      Re: Degree of freedom - proc mixed
Comments: To: che@JOUY.INRA.FR
In-Reply-To:  <3843DCE5.1F85@jouy.inra.fr>

Date sent:              Tue, 30 Nov 1999 15:19:17 +0100
Send reply to:          Christelle Hennequet-Antier <che@JOUY.INRA.FR>
From:                   Christelle Hennequet-Antier <che@JOUY.INRA.FR>
Organization:           INRA
Subject:                Degree of freedom - proc mixed
To:                     SAS-L@LISTSERV.UGA.EDU


Christelle,

Depending on the syntax, you may be specifying the
Sattethwaite's Procedure to approximate the degrees of freedom.
See Neter et al. (1996) Applied Linear Statistical Models.

This iterative method may result in non-integer values.

-Bob

> I 'd like to have more information about degree of freedom. In
> particular, why sometimes there aren't integer? How there are
> calculated?
>
> Thanks.
> --
>  Christelle