Deviance

Deviance is defined as the difference in -2log(Likelihood) of the current model and -2log(Likelihood) of the saturated model. The saturated model is the model with the number of parameters equal to the sample size. Deviance divided by its degrees of freedom is an estimate of over-dispersion, or c-hat.

For some data types, there are issues about how to compute the value for the saturated model. In these cases, you might want to include the -2log(L) quantity in the Results Browser by setting this option in the File | Preferences window.