I would like to specify a two-level logistic regression model with random intercept and random slope.
Dependent variable: hospitalization (1) or no-hospitalization (0).
Independent variables: age, number of drugs used, comorbidity, others...
Multilevel structure: patients clustered within hospitals. Hospitalization rate varies across different hospitals.
The goal is to identify variables in the model that explain the largest part of variation in hospitalization rate.
I have two questions:
Can I run several models each time allowing for random intercept + random slope including all other variables as fixed. I would like to have a look at the size of variance that is explained by each patients' characteristic, but when i add several random slopes model does not converge.
When I specify the random slope then the intercept variance is inflated. How can I interpret that?