missing data
Posted: Tue Nov 13, 2012 10:02 am
Dear colleagues,
I'm running a complex three level model (25 countries, 1344 schools and 57771 individuals) on problematic alcohol use, individual control variables and risk factors and structural country variables (so only variables on individual and country level). Because of complexity they advised us to run it in Bayes. With runmlwin it didn't succeed till now (I mail with this forum also) but with the more time consuming MLWIN Bayes it does. Although I thought one of the advantages of Bayes is its handling with missing data it shows that at the end the model runs with only 55% of the cases.
The new Mplus-version which works with three levels has the opportunity to impute X-variables meanwhile (But I don't have this new upgrade). Is there such a opportunity also in MLwin? Or should I first impute the variables with Realcome and that analyse the data set.
Thanks for your suggestions.
Big best, Harrie
I'm running a complex three level model (25 countries, 1344 schools and 57771 individuals) on problematic alcohol use, individual control variables and risk factors and structural country variables (so only variables on individual and country level). Because of complexity they advised us to run it in Bayes. With runmlwin it didn't succeed till now (I mail with this forum also) but with the more time consuming MLWIN Bayes it does. Although I thought one of the advantages of Bayes is its handling with missing data it shows that at the end the model runs with only 55% of the cases.
The new Mplus-version which works with three levels has the opportunity to impute X-variables meanwhile (But I don't have this new upgrade). Is there such a opportunity also in MLwin? Or should I first impute the variables with Realcome and that analyse the data set.
Thanks for your suggestions.
Big best, Harrie