# Calculate goodness-of-fit (with deviance) to compare averaged models?

I need to compare the goodness of fit of several averaged logistic regression models by calculating the deviance explained. I'm using the MuMIn package in R to average many logistic regression models into a single averaged model. I ultimately want to compare the explanatory power of several averaged models, in part by using the deviance explained.

My questions are:

1. Does deviance explained apply to averaged models as a strong measure of the goodness of fit?

2. How does one calculate the deviance explained (calculated as the null deviance less the residual deviance as a proportion of the null deviance) from the averaged model output from the model.avg() command in MuMIn?

Examining the structure of the averaged model object indicates that the null and residual deviances are calculated on each individual model that contributes to the averaged model, but I'm not sure how to extract them and then calculate the overall deviance explained by the averaged model.