R help - Sep 2012 - appropriate test in glm when the family is Gamma

Dear R users,

Which test is most appropriate in glm when the family is Gamma?
 
  In the help page of anova.glm, I found the following
?For models with known dispersion (e.g., binomial and Poisson fits) the
chi-squared test is most appropriate, and for those with dispersion estimated by
moments (e.g., gaussian, quasibinomial and quasipoisson fits) the F test is most
appropriate.?
 
My questions :
(1)When the family is Gamma, is F-test most appropriate? or chi-square-test? or
another?
(2)Aside from the tests made by anova.glm, there are other tests (e.g. Wald and
score). Are they ( or one of them) more appropriate when the family is Gamma?

?? ?? <ekasuscb <at> kyushu-u.org> writes:

> Which test is most appropriate in glm when the family is Gamma?  
> In the help page of anova.glm, I found the following ?For models
> with known dispersion (e.g., binomial and Poisson fits) the
> chi-squared test is most appropriate, and for those with dispersion
> estimated by moments (e.g., gaussian, quasibinomial and quasipoisson
> fits) the F test is most appropriate.?
 
> My questions :
> (1)When the family is Gamma, is F-test most appropriate? 
>  or chi-square-test? or another?
  F is most appropriate.  In the base glm function, I believe that
the dispersion parameter is indeed estimated by moments (there is
a gamma.shape() function in MASS that refines the estimate via MLE).
> (2)Aside from the tests made by anova.glm, there are other tests
> (e.g. Wald and score). Are they ( or one of them) more appropriate
> when the family is Gamma?
  Wald test would probably be less accurate (you get Wald t
tests on the individual parameters from summary(), you could
get Wald F tests on factors from car::Anova) ; score test is
fairly rarely implemented in R, as far as I can tell ...