Hi
As far as I know that assumptions of likelihood ratio test is violated under the survey design , so we can not use deviance to select the model . I have a logistic model under complex survey design and I my question is how to compare my models.
Thanks