P values option

Martin Maechler Martin Maechler <maechler@stat.math.ethz.ch>
Wed, 14 Oct 1998 10:43:50 +0200

(From original R-help, then  R-core discussion;
 this is potentially of interest to more people. MM)

    JL>  2. meaningless P-values and stars by default in summary(glm())

    TL> The default of stars is easily changed in a site-level profile,
    TL> which can be given to students if you are worried about their minds
    TL> being contaminated by stars. I'm going to do this on the
    TL> installation here.
    TL> Perhaps we should have an option to leave p-values off summary.{lm,glm}?

    KH> I thought we had this.  print.coefmat() uses .Options$signif.stars ...
    PD> Note: *p-values*, not stars. Not a good idea, in my opinion.
nor in mine.  

    TL> I agree it's a bad idea to leave p-values off, but I still think
    TL> it's reasonable to have *an option* to leave them off. Some people
    TL> don't like Wald tests in non-linear models (even in generalised
    TL> linear ones), some don't like giving p-values for individual
    TL> components of factors, other people believe in the likelihood
    TL> principle and don't want p-values at all.  This sort of
    TL> customisation would presumably just be for teaching purposes.

Ok.  Given the  print.coefmat(.) interface,
it would be pretty easy to implement yet another option, let's say
	options(show.coef.Pvalues = TRUE)
		better name ??

If set to FALSE, it would not print P-values with print.summary.g?lm(.)
but it would print  for  print.anova
	(Here we are talking about  R 0.63, 
	 where print.tabular is not used anymore, see
	 CRAN src/devel/R-devel.tar.gz)


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