c-hat when deviance = 0

[mcobben]

Good day! I'm trying to analyse roe deer data (montly radio contact for 40 months, using the known fate model with Kaplan-Meier input for 2 groups, male and female. Survival estimates are high, with a lot of boundary values).
Fitting the saturated model I find the deviance is very small (0,4E-11) and c-hat is (thus) very small as well. 1. Can this deviance be correct? 2. If so, is it because of [quote] Deviance is defined as the difference between -2log Likelihood for the model of interest and the -2log Likelihood of the saturated model. Program MARK -- Gary C. White. 3. Again if so, how do can you estimate c for the saturated model? Is c-hat=1 because the average of the simulated c-hats (=0) equals the observed c-hat (=0)? Or is c-hat just << 1, indicating (extremely) underdispersed data? Or is there a reason why you shouldn't have to estimate c for the saturated model?
In other words: please help!

Thanks,
Marleen[/quote]