@Pablo: I have a general question regarding your analytical approach of ordinal multilevel models:
How do you deal with the scaling-problems of nested multilevel ordinal models? By nested I mean stepwise entering variable blocks to finally end up with an full model? This approach is very common in my discipline to check how the coefficients of the variables and the variance components of Level 1 and Level 2 behave after entering further variables into a (multilevel) model. In linear models this not a problem, but in non-linear models, however, this is not feasible. For reference see:
Did you have any idea how to deal with this situation? How do you do it?
How do you deal with the scaling-problems of nested multilevel ordinal models? By nested I mean stepwise entering variable blocks to finally end up with an full model? This approach is very common in my discipline to check how the coefficients of the variables and the variance components of Level 1 and Level 2 behave after entering further variables into a (multilevel) model. In linear models this not a problem, but in non-linear models, however, this is not feasible. For reference see:
- Karlson, K. & Holm, A. & Breen, R. (2012). Comparing Regression Coefficients Between Same-Sample Nested Modes Using Logit and Probit: A new Method.
- Breen, R. & Karlson, K. & Holm, A. (2013). Total, Direct, and Indirect Effects in Logit and Probit Models
Did you have any idea how to deal with this situation? How do you do it?
Comment