Hi Chris,

There is no way to get those covariance matrices from R directly, but you could calculate them from the imputed datasets and then average the quantities together using the Rubin rules to produce estimates of the covariance matrices and means. These imputed datasets are in the original scales, so this should produce what you are looking for.

Cheers,
Matt

On Fri, Aug 14, 2015 at 4:39 AM Chris McManus <i.mcmanus@ucl.ac.uk> wrote:

Hi -- I am new to this list, have tried searching for an answer on

the existing posts, and don't think there is one already, but

apologies if I have missed it.

My primary interest is in the estimated means and covariances from

the EM procedure (and bootstrapping helps as I need standard errors

of means, covariances, etc).  It is clear that covMatrices and mu are

not on the original scale, and the JSS paper says on page 36 that

"they refer to the variables after being transformed, centered and scaled".

The question is therefore, How do it get them on the original scale?

Is there a transformation matrix somewhere? Do I just alter the means

and SDs back to the original  raw values, or has there been a

rotation of some sort as well?  For my purposes I do need to know the

values on the raw scale measures.  I guess I am missing something

here but am not sure what it is.

Many thanks!

Chris McManus

University College London

--

Amelia mailing list served by HUIT

[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=amelia

More info about Amelia: http://gking.harvard.edu/amelia

Amelia mailing list

Amelia@lists.gking.harvard.edu

To unsubscribe from this list or get other information:

https://lists.gking.harvard.edu/mailman/listinfo/amelia