AIC() and anova() for svy_vgam models

[coreysparks]

Thanks for porting the VGAM models to use your survey package. Any plans on allowing survey::AIC() or survey::anova() to accept these svy_vglm() models?

[tslumley]

I'll put it on the list. AIC will be a bit more tricky than anova, because it's necessary to worry about the influence functions for nuisance parameters (like the dispersion parameter in a Normal linear model), but that shouldn't be impossible. Anova for symbolically nested models should be pretty straightforward (and thus regTermTest as well)

[symbalex]

Seconded on AIC, if you get the chance! Thank you :-)