(De)selecting multivariate mode (!ASUV !ASMV)

!ASUV

!ASUV means univariate analysis is required although the data is presented in a multivariate form. Specifically, it allows you to have an error variance other than IDENTITY by US. If there are missing values in the data, include mv in the fixed part of the linear model. The intrinsic factor Trait is defined and may be used in the model.

This option is used for repeated measures analysis when the error variance structure required is not the standard multivariate unstructured matrix.

!ASMV t

!ASMV n indicates multivariate analysis is required although the data is presented in a univariate form. 'Multivariate Analysis' is used in the narrow sense where an unstructured error variance matrix is fitted across traits, records are independent, and observations may be 'missing' for particular traits, that is, the data record must be present but the data value may be NA.

The data is presumed arranged in lots of n records where n is the number of traits. It may be necessary to expand the data file to achieve this structure, inserting a missing value NA on the additional records. This option is sometimes relevant for some forms of repeated measures analysis. There will need to be a factor in the data to code for trait as the intrinsic Trait factor is undefined when the data is presented in a univariate manner.

Return to start