For some background: my dissertation is on the factors that affect successful rehabilitation for casualties within a wildlife hospital. I have several variables including age, code of casualty, taxonomic group, etc.
Most of the variables are categorical, with the exception of time. The outcome is died/survived, so I've run a GLMM with a logit function for all tests, like the example below:
analysis<-glm(Resultfac~Agefac + Codefac + Species.Typefac, family = binomial, data = dataset)
My supervising professor suggested I remove terms to test for "overall significance" and that I should use the ANOVA function with a chi-square test to do so:
significance<-update(analysis,~.-Species.Typefac)
anova(analysis, significance, test = "Chisq")
For the majority of these tests, I'm getting outputs where p<0.05 (in fact the majority are p=2e-16), which indicates significance. However, when reading other papers that include model outputs, many list chi-square values that are >0.05. Even one of my professor's papers includes chi-square outputs that are all >0.05. After doing some research I've learned that having a chi-square value of larger than >0.05 is indicative of an individual term, in which case none of my terms would be individual.
Ultimately, I'm confused as to if my terms are significant or not when the ANOVA chi-square tests have an output of <0.05. I would very much appreciate some clarification so that I can make sense of my results, and what factors are ultimately significant.
