Hi Scholars,

I have a dataset that overrepresented respondents in some of important socio-demographic variables (retired respondents, females, older in age, and those from low income). I have done latent profile analysis(LPA) using items related to general consumer behaviours and then tried to assess variables (covariates) predicting class membership (multinomial regression). So my questions are:

1. Can I include more than one weighted independent variables at a time conducting multinomial regression?
2. Do I need to use weighted indicator variables (general consumer behavioural items) to do LPA in the first place assuming the response might be already skewed as socio-demographic variables skewed)? If so, please advice me how to use weighted variables for LPA and multinomial regression.

Great appreciations for your help!

Liyu