Multiple Imputation Method - Including complex weights

#1
Hello, I was trying to perform a multiple imputation on missing data of my project. However, I was not able to find how to include complex weights into the analysis. I'm assuming that it would be added to the syntax. Sorry for the request but if it is so, would you be able to inform me what the code is and where to put it into this code? Thank you.

*Impute Missing Data Values.
DATASET DECLARE Dataset1.
DATASET DECLARE Iterations.
MULTIPLE IMPUTATION [REDACTED]
/IMPUTE METHOD=AUTO NIMPUTATIONS=5 MAXPCTMISSING=NONE MAXCASEDRAWS=50 MAXPARAMDRAWS=2
/CONSTRAINTS [REDACTED]
/MISSINGSUMMARIES NONE
/IMPUTATIONSUMMARIES MODELS DESCRIPTIVES
/OUTFILE IMPUTATIONS=Dataset1 FCSITERATIONS=Iterations .
 

Lazar

Phineas Packard
#2
I doubt it is possible. MI is a Bayesian analysis and there is no clear way to incorporate weights into this. Your best bet is to include variables into your MI on which the weights are based. e.g., strata etc. though you may need to compromise a little given the MI procedure in SPSS often has issues with categorical variables.

Depending on the missing mechanism another choice might be to generate attrition weights and just use ML for any remaining missing. This is the approach I typically take with longitudinal data.