Page 1 of 1

Errors in REALCOM-IMPUTE -- NaN, Empty Matrix

Posted: Fri Mar 15, 2013 8:57 pm
by uthResearcher
I have been trying to use REALCOM-IMPUTE for the first time, and I can't get it to run the model. I've installed two different MATLAB runtime environments:

1. When I install the one listed here http://www.bristol.ac.uk/cmm/software/r ... #downloads, the program doesn't even run due to a missing mclmcrrt710.dll file.
2. Somewhere else, I was told to request a link to the MATLAB runtime for REALCOM, which I did. The link didn't work (can't post it here, by their request), but looking at the url, it seemed to iterate by ones so I changed from a '2' at the end to a '3', and it downloaded a file called matlab-realcom so I figured that must be right (now I'm not so sure). Anyway, the program ran so I continued on.

I'm a SAS user, and I was able to create a file that the program reads correctly. It recognizes the variables (1 continuous response, 1 continuous explanatory for my test run), the number of observations and the number of level 2 units, and the equations are correct. However, when I click "Start MCMC run," nothing happens. So I switched over to the command prompt for information, and I keep getting the same error(s):
REALCOM-IMPUTE_error.jpg
REALCOM-IMPUTE_error.jpg (60.08 KiB) Viewed 5131 times
It seems to be failing to create the BETASTART_2 matrix correctly, which results in a Not a Number (NaN) for the SIGMASTART matrix. I've tried many models, with several response variables, and checked off the list of troubleshooting tips in Goldstein (2011) REALCOM-IMPUTE: Multiple imputation using MLwin. I feel like I've tried everything and I'm not sure what to do next. Any help would be appreciated, and thanks in advance.

EDIT: words.
EDIT 2: I heard back from the person who sent me the link to the MATLAB installer, and I did use the correct one.

Re: Errors in REALCOM-IMPUTE -- NaN, Empty Matrix

Posted: Mon Mar 18, 2013 12:34 pm
by ChrisCharlton
Would it be possible for you to send the files that you are having problems with to Professor Goldstein (contact details here: http://www.bris.ac.uk/cmm/team/hg/) and he will try to determine what the problem is?