Large SEs after imputation

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
gilleecl
Posts: 27
Joined: Fri Dec 17, 2010 11:24 am

Large SEs after imputation

Post by gilleecl »

Hi,

I'm just getting started with Realcom Impute and I wonder if you could help me please. My model includes one variable with missing values and 6 (including dependent var) with no missing vals. Prior to running multiple imputation, it looks ok in MlwiN; i.e. parameter estimates and SEs look reasonable. Once I run multiple imputation in Realcom Impute, the SEs for the variables with no missing values become really huge and the parameter estimate and SE for the variable with imputed values are both 0.000. I think it maybe that I'm doing something simple incorrectly; i.e. when I click imputation -> start analysis; things look ok, values look reasonable, but they are in blue. Once I press start or more, PE and SE for variable with imputed values become zero and SEs for other variables get very very large. Is there something else I should do at this point rather than clicking start/more?

Many thanks,
Lorraine
ChrisCharlton
Posts: 1252
Joined: Mon Oct 19, 2009 10:34 am

Re: Large SEs after imputation

Post by ChrisCharlton »

Clicking:

imputation -> start analysis

is the correct way to perform the analysis on your set of imputed data. It doesn't turn green because the results from several model fits have been combined and therefore don't technically match the last remembered iteration. If you press "start" then MLwiN will attempt to fit the model using the data that contains the missing values, hence the strange values that you were seeing. For an example of a complete imputation analysis see:

http://www.bris.ac.uk/cmm/software/real ... tation.pdf
gilleecl
Posts: 27
Joined: Fri Dec 17, 2010 11:24 am

Re: Large SEs after imputation

Post by gilleecl »

great, thanks.
Post Reply