[Rd] t.test (PR#1086)

Prof Brian D Ripley ripley@stats.ox.ac.uk
Sat, 8 Sep 2001 07:32:28 +0100 (BST)

On Sat, 8 Sep 2001 menstav@hotmail.com wrote:

> Full_Name: Menelaos Stavrinides
> Version: 1.3. 1
> OS: Windows 98
> Submission from: (NULL) (
> When model simplification is used in glm (binomial errors) and anova is used two
> compare two competitive models one can use either an "F" or a "Chi" test.
> R always performs an F test (Although when test="Chi" the test is labeled as
> Chi, there isn't any difference between F and Chi test)
> Let us assume model1 and model2, where model2 a simplified version of model1
> model1_glm(y~factor1*factor2,binomial)
> model2_update(model1,~.-factor2)
> anova(model1,model2,test="F")# will give the same result as
> anova(model1,model2,test="Chi")

See help(family) and help(anova.glm).  For a binomial family the scale
estimate in the F test is fixed at one, so they should be the same.
If you want extimates scale, use a quasibinomial family.

This applies to PR#1087 too.

Please do read the section of BUGS in the FAQ.

Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch