Missing values in explanatory variables
!MVINCLUDE or !MVREMOVE
ASReml will abort the analysis if it finds missing values in the design matrix
causes the missing value to be treated as a zero.
causes ASReml to discard the whole record.
1 Identify which model terms have the missing values from the data summary,
2 Understand why observations are missing,
3 Decide whether to delete the records, centre covariables,
impute a value to the missing observations
or treat missing treatment levels as an extra category.
Use !MVREMOVE if you are not sure that !MVINCLUDE is appropriate.
is an easy way to discard records with missing explanatory variables,
if it causes the job to fail for other reasons, it may be
necessary to explicitly drop records containing
missing values in particular fields with the
transformation can change missing values to a nominated value.
Treating missing values as
in covariates is usually only sensible if the covariate is
(has mean of zero).
the factor level is zero (or missing and the
qualifier is specified), no level is assigned to the
for that record.
Return to start