GOF tests & individual covariates

[tds]

Hello,

Can you please verify that there is currently no way to do a GOF test on a model with individual covariates?

Assuming this is true, if my most saturated model w/o individual covariates is phi(.)p(.), is it even possible or reasonable to do a GOF test? (I've tried just to see what happens, and MARK won't run a bootstrap or median c-hat GOF on the dot model. I'm assuming it's because it's not possible to run a GOF test on a dot model?)

If I can't run a GOF test, how can I evaluate the fit of my saturated model?

Much thanks

[bacollier]

Hello,

Can you please verify that there is currently no way to do a GOF test on a model with individual covariates?

Chapter 11, section 11.9, verifies it for you.

Assuming this is true, if my most saturated model w/o individual covariates is phi(.)p(.), is it even possible or reasonable to do a GOF test? (I've tried just to see what happens, and MARK won't run a bootstrap or median c-hat GOF on the dot model. I'm assuming it's because it's not possible to run a GOF test on a dot model?)

If I can't run a GOF test, how can I evaluate the fit of my saturated model?

Much thanks

The most saturated model would likely be a phi(t)p(t), what it looks like you are showing is that your best model without covariates is a dot model, am I interpreting that correctly? If so, check the fit of the most saturated model in your set (which should be phi(t)p(t) or some variant therein, not the dot-dot model you show), which does not have to be the best model per AIC rank last I checked.

\bret