Specially written REALCOM macros with a menu interface are now available for handling quite general missing data patterns. These will deal properly with categorical as well as normal data and also with multilevel structures. The model of interest is first set up in MLwiN in the usual way and then the variables exported to REALCOM-Impute and then the imputed datasets returned to MLwiN where they will be fitted and combined automatically for the specified model of interest.
- Download REALCOM Impute from our REALCOM download page. If you have any difficulties receiving your download, please email firstname.lastname@example.org. Note: You will need MATLAB runtime installer (more >>)
- * A new version of REALCOM-IMPUTE can now be downloaded. Some bugs have been fixed. The old version sometimes crashed or gave clearly incorrect results when more than one categorical response was present at either level 1 or level 2. This has now been updated with new bug fix for models with level 2 responses.
- 23 April - A bug causing some models to crash realcom-impute has been fixed.
- 19 June - A new bug has now been fixed. This caused errors when missing values for level 2 responses were imputed.
- MATLAB runtime installer for Realcom - You will need this if you do not have MATLAB (www.mathworks.com) already installed. This is available from the Mathworks MATLAB Compiler download page. Download the R2012b Windows 32-bit version.
- Goldstein H. (2009) REALCOM-Impute: Multiple Imputation using MLwiN, User Guide (PDF, 0.2 mb, 12-Jul-11)
REALCOM Impute and Stata
- Data can now be exported to REALCOM-Impute directly from Stata. These commands allow you to prepare input data for REALCOM-Impute in Stata, and then load the results back into the appropriate Stata structures. Further details
- Realcom user forum - a dedicated Realcom section within the MLwiN user forum.
Previous Bugs (earlier versions of Realcom-Imputation)
To resolve the following bugs please ensure you have the latest version of REALCOM-Impute
- A bug that affected single level models has also been fixed.
- A bug that can affect the covariance estimates for discrete responses has been fixed (22-Mar-11)
- If you try to fit a single level imputation by removing the level 2 identifier, the program will crash. This will be fixed as soon as possible.
- Note: When retrieving and running imputed datasets in MLwiN you should first issue the following command in the command interface window: maxi 100. If you do not do this the number of interations for each set of estimates will be the number that happens to be specified in your worksheet and this may be too small. This was corrected in later releases of MLwiN.
Missing data are a persistent problem in social and other datasets. A standard technique for handling missing values efficiently is known as multiple imputation and the software REALCOM-Impute is unique in that it has been designed to implement this procedure for 2-level data. Apart from being able to deal with 2-level data it can also handle properly categorical data, whether in the response or predictor variables in a model. An interface is provided with MLwiN that allows users to carry out the full procedure and fit their final model semi-automatically.
Note: some of the documents on this page are in PDF format. In order to view a PDF you will need Adobe Acrobat Reader