[vsni.co.uk] Contact us
Author Message
Post new topic

  ASReml  ~  missing values in the response variable, multivariate model

vincent_careau
Posted: Wed Oct 31, 2018 12:15 pm Reply with quote
Joined: 02 Oct 2018 Posts: 3
Hi,
I have a question regarding the estimation of missing values in the response variables of a multivariate mixed model. In the asreml-R user manual, we can read that:
“Records with missing values in the response are included by default (na.method.Y = ”include”) and estimated as a consequence of fitting the model. A factor labelled mv is created and included in the sparse equations, and the solutions are returned in coef(object)$sparse. An alternative action is ”omit” which excludes units with missing values in the response. Missing values must be estimated in a multivariate analysis.”

I’m having a hard explain what is done exactly in the multivariate mixed model that has missing values in the response variables. Looking at the “coef(object)$sparse” we can see there is a list of “mv” values. So I understand these are estimated values. But how were these estimated? Is this a case of “data augmentation”? (also referred to as “full information maximum likelihood” in the stats literature). Is it OK to describe the procedure as a case of “model-based data imputation”?
View user's profile Send private message
Arthur
Posted: Fri Nov 02, 2018 1:06 am Reply with quote
Joined: 05 Aug 2008 Posts: 471 Location: Orange, NSW
Dear Vincent,

When mv is included on the model, ASReml sets the missing response to Zero and creates a design matrix column in 'mv' of -1. Assuming other design terms are set, this equation becomes like

0 = XB - Ym
or Ym=XB

so the reported missing value is whatever is predicted from the rest of the model (the data that is present).

In standalone ASReml, missing values in multivariate response variables are handled by using a unit
specific version of R which omits the rows where the response is missing.

_________________
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)
View user's profile Send private message Send e-mail Visit poster's website

Display posts from previous:  

All times are GMT
Page 1 of 1
Post new topic

Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum