[19092] in s-news-athena
RE: [S] GLM, quasi likelihood , restriction to dispersion > 1
daemon@ATHENA.MIT.EDU (Spyros Vrontos)
Wed Sep 1 20:53:52 1999
Message-Id: <3.0.3.32.19990902034304.006f143c@stat-athens.aueb.gr>
Date: Thu, 02 Sep 1999 03:43:04 +0300
To: s-news@wubios.wustl.edu
From: Spyros Vrontos <svrontos@stat-athens.aueb.gr>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Hi
I had asked how one could restrict dispersion >1.
for example:
summary(test, dispersion > 1)
does not work.
I would like to thank Stephen Smith for giving the following answer:
>Dispersion has to be set to a specific value: See section from help file
>for summary.glm below.
>
>dispersion a supplied value for the dispersion parameter. The default
>is 1 for the binomial or poisson families. For other families the default is
>the residual Chi-square statistic divided by the residual degrees of
>freedom. For the Gaussian family, for example, the dispersion parameter is
>the error variance and the default is the Residual sum of squares divided by
>the residual degrees of freedom. Supplying a value of 0 causes the
>Chi-squared estimate to be used in all cases.
>
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
