2 level vs. single level imputation model

Welcome to the forum for REALCOM users. Feel free to post your question about REALCOM here. The Centre for Multilevel Modelling take no responsibility for the accuracy of these posts, we are unable to monitor them closely. Do go ahead and post your question and thank you in advance if you find the time to post any answers!

Go REALCOM (Developing multilevel models for REAListically COMplex social science data) >> http://www.bristol.ac.uk/cmm/software/realcom/
Post Reply
zahra777
Posts: 15
Joined: Mon Feb 03, 2014 1:39 am

2 level vs. single level imputation model

Post by zahra777 » Mon Mar 03, 2014 9:17 am

Dear experts,

Can I have a 2 level imputation model where I predict only covariates at Level 2 without using my level 1 variable ?

I have a 2 level repeated measure data structure. My level 1 variable is my dependent variable and it has missing values.
If I do not use my dependent variable ( level 1) to predict or be predicted by the level 2 variables, can I still have a 2 level imputation model or I am bound to have a single level imputation model? Can I still account for the multilevel structure of my data in my imputation model?

Many thanks
Zahra

Harvey Goldstein
Posts: 49
Joined: Sun Sep 06, 2009 5:30 pm

Re: 2 level vs. single level imputation model

Post by Harvey Goldstein » Mon Mar 03, 2014 11:57 am

If your model of interest is 2 level and you have missing values in the response then when imputing these you should run a 2 level imputation model.

zahra777
Posts: 15
Joined: Mon Feb 03, 2014 1:39 am

Re: 2 level vs. single level imputation model

Post by zahra777 » Mon Mar 03, 2014 1:50 pm

Thank you very much for your reply.

So, just to confirm, even though there is no level 1 variable in my imputation model, and all I am predicting is level 2 covariates, I can still use 2 level imputation model?

Many many thanks

Zahra

Harvey Goldstein
Posts: 49
Joined: Sun Sep 06, 2009 5:30 pm

Re: 2 level vs. single level imputation model

Post by Harvey Goldstein » Mon Mar 03, 2014 2:00 pm

Strictly speaking you should set up a 2-level imputation model where any level 1 vbles in the MOI are responses (at level 1 in the imputation model).
Harvey Goldstein

zahra777
Posts: 15
Joined: Mon Feb 03, 2014 1:39 am

Re: 2 level vs. single level imputation model

Post by zahra777 » Mon Mar 03, 2014 2:39 pm

Thank you, but what if I can't use my level 1 variable in my imputation model?

My level 1 variable has missing values so it cannot be used as auxiliary variable.
My level 1 variable is my dependent variable and according to some school of thoughts, dependent variables should not be predicted in imputations.

How can I account for the multilevel structure of MOI when doing multiple imputation?

Greatly appreciate your reply.

Best regards,
Zahra

Harvey Goldstein
Posts: 49
Joined: Sun Sep 06, 2009 5:30 pm

Re: 2 level vs. single level imputation model

Post by Harvey Goldstein » Mon Mar 03, 2014 2:52 pm

All variables in the MOI, whether responses or predictors do need to be in the imputation model.

zahra777
Posts: 15
Joined: Mon Feb 03, 2014 1:39 am

Re: 2 level vs. single level imputation model

Post by zahra777 » Mon Mar 03, 2014 2:59 pm

sorry for so many questions and thank you very much for replying,

So in other word, if I am using Realcom for multiple imputation, I have to use all my variables in MOI ( dependent and predictor variables) either as auxiliary or response?

And in my 2 level data structure, if I am not planning to predict missing values of my dependent variable in my imputation model, do I have to use single level multiple imputation? ( i.e. just at level 2? ) or even that would be wrong because my MOI has 2 levels?

Many thanks
Zahra

Post Reply