When I run nnet::mutinom, I can get the p-value for a term via Anova(model, type=c('III')) etc. But this method doesn't work for svy_vglm objects. I get the error:
Error in x$terms %||% attr(x, "terms") %||% stop("no terms component nor attribute") :
no terms component nor attribute
There has to be a work around or a way to get type 3 analysis of effects for multinomial survey regression.
This is NOT a duplicate question, I saw this solution:
beta<-coef(model)[index]
V<-vcov(model)[index,index]
teststat <- crossprod(beta, solve(V,beta))
pf(teststat, df1=length(beta), df2=degf(design), lower.tail=FALSE)
which is great but I have some 300 terms so it's not efficient and I don't think computes type III.
Any help would be appreciated