Hello everyone,

I am working with a database that have information on a survey asking for mosquito nets maintenance and use. I have information both at the household level and the individual level. Our outcome of interest is the individual use of mosquito nets. Our exposures are both variables that are identical for all household members (such as type of hosusehold) and variables that are unique for each individual (such as sex, age, studying status or employment status).

I want to do a table with the result of the cross-tabulation of the different exposures and our main outcome (in %), as well as adding the crude OR for each variable.

For those exposures identical between household members, I need to use random effects logistic regression (xtlogit). However, what I need to use for those variables different between household members? Simple fixed effect logistic regression? Or do I need to take into account "household level" as well mandatorily (e.g. maybe as a counfounder). Even if exposures are variables that have different options depending on the individual, maybe members of the same household are more alike (e.g. more prone to study all or to be employed all?).

Please note that at this point I do not want to adjust for more variables nor combine both types of exposures (where a mixed effect model will be necessary).

Thank you very much!