For GLM you could do:

> model1 anova(model1, test="F") # F test appropriate for e.g. the following values for family: "gaussian", "quasibinomial", "quasipoisson".

For gam:

> model1 anova(model1, test="F")

Regarding the inclusion of interaction terms in a GAM-model, there is e.g. the following webpage: http://stats.stackexchange.com/questions/32730/how-to-include-an-interaction-term-in-gam

Regards, G. Haege

For GLM you could do:

> model1 anova(model1, test="F") # F test appropriate for e.g. the following values for family: "gaussian", "quasibinomial", "quasipoisson".

For gam:

> model1 anova(model1, test="F")

Regarding the inclusion of interaction terms in a GAM-model, there is e.g. the following webpage: http://stats.stackexchange.com/questions/32730/how-to-include-an-interaction-term-in-gam

Regards, G. Haege

Hi there, 

The anova() function will do what you want.  Make sure you pick the right test when using the "test = " argument.

Best,   Jorge Velez.-

Dear Jorge and Gisela,

Thanks for your guidance. I used anova() function mainly for model comparison specially in model simplification but i was not sure can i replace it with summary() function. 

@ Heage are you sure we can apply "test=F" for quisipoisson and quisibinomial as well? I think "test=Chi" would fit to this families.

Warm regards,

Mehdi

According to webpage https://stat.ethz.ch/R-manual/R-devel/library/stats/html/anova.glm.html, "test=F" is appropriate for models "with dispersion estimated by moments (e.g., gaussian, quasibinomial and quasipoisson fits)". "For models with known dispersion (e.g., binomial and Poisson fits) the chi-squared test is most appropriate".

Regards, G. Haege

Dear Gisela,

Thanks for your references.do you know any references to sum up all possible "test" for GLM? Could we apply them for GLMM as well? 

Warm regards, Mehdi

Hi,

I have a similar question for glmer. How to get only main factors and interactions without pairwise comparisons of their levels in glmer?

Thanks!