Multiple imputation of multilevel data

[angranli]

Hi everyone, I am working on a multilevel data which students are nested within schools. Both individual level and school level have missing values. My general question is how to impute variables with missing values at different level? Thank you so much!

[angranli]

Could anyone help?

[angranli]

I will make my question more clear. My major concern is the missing data in level-2. We know that level-2 predictors should assign the same imputed value to all members within the same school. Some people suggest creating two data sets, one with only individual-level data, and one with school-level data, and do separate imputations within each data set while using the results from one in the other. However, if I do this, for example, generating 5 imputed date sets for individual-level, and 5 imputed data sets for school-level. How could I merge them together? There are 5*5 (25) different combinations, right? How can we address this issue?

Thank you so much!

[Rich Goldstein]

have you looked at example 4 in the MI manual (on "Panel data and multilevel models")?

have you looked at the Stata FAQ “How can I account for clustering when creating
imputations with mi impute?”; see http://www.stata.com/support/faqs/stat/impute cluster.html

[angranli]

Hi Rich,

Thank you so much for your reply!

For the MI manual "Panel data and multilevel models", could you give me more information about it? Thanks again!

I read the Stata FAQ, but they don't mention about how to impute level-2 variables. Could you give me more insights for this?

I am really appreciated your help!!

Best,
Angran

[Rich Goldstein]

I don't understand your first question (re: the manual) - did you look at the manual? what more info could I supply; note that I referred to the manual for version 14; if you hare using some other version, then, as the FAQ states, you need to tell us

for some reason I can't currently access the FAQ (though I did yesterday) so I can't answer that either - sorry