Page 1 of 1

Running 2-level MLM from variance-covariance matrices

Posted: Thu Aug 25, 2016 11:34 am
by mattycourtney
Hi All,

My first time posting a question. I have been granted access to a governmental database (with individual-level public data) and can only work with 'raw data' inside this remote government controlled environment. I need to carry out 2 or 3 level MLM and only SAS is made available to me inside this environment. However, I would much prefer to use MLWin on my office desktop as I have more experience with it.

I am able to get clearance for variance-covariance matrices from this government body as such data don't contain personal information.

So, is there any way to carry out MLM with MLWin by reading in variance-covariance matrices? I can do factor analysis in SPSS based on such data.

Sorry if this seems like a silly question; I'm relatively new to MLM etc.

Any advice would be appreciated.

Cheers,

Matt

Re: Running 2-level MLM from variance-covariance matrices

Posted: Fri Sep 02, 2016 10:41 am
by GeorgeLeckie
Hi Matt,

I'm afraid, in MLwiN and all other standard multilevel modelling data you need to fit models to the the underlying data, not to the summary statistics derived from these data.

Suggest you encourage the Government department to install MLwiN on their computers.

Best wishes

George