Dear Statalist-Users,
I am using the multiple imputation command "mi".
I also created a weight-variable based on official statistics combining information for gender and education.
I was wondering whether:
1. it is recommended to combine weighting and mi (Stata allows doing so, but might there be good reasons to not do it?)?
2. the weight-command should be used for mi impute chained var1-var5 [pweight=weightvar] AND for mi estimate: ologit var1 var2-var5 OR in only one of the commands?
3. it is required or mandatory to use the svy: prefix? If required/mandatory: why? In my case, there is no survey structure variable such as strata.
Thank you very much,
Seb
I am using the multiple imputation command "mi".
I also created a weight-variable based on official statistics combining information for gender and education.
I was wondering whether:
1. it is recommended to combine weighting and mi (Stata allows doing so, but might there be good reasons to not do it?)?
2. the weight-command should be used for mi impute chained var1-var5 [pweight=weightvar] AND for mi estimate: ologit var1 var2-var5 OR in only one of the commands?
3. it is required or mandatory to use the svy: prefix? If required/mandatory: why? In my case, there is no survey structure variable such as strata.
Thank you very much,
Seb
