1. Home
  2. Linear Mixed Models (REML) – Correlated Errors

Linear Mixed Models (REML) – Correlated Errors

Defines a covariance structure for random effects in a linear mixed model fitted by REML.

Correlated error models

Lists the random model terms to which a covariance structure is to be applied. Select the random model term that you want.


Specifies whether the covariance structure is to be defined as a single structure for the whole term, or if it is to be constructed as a direct product of matrices corresponding to the factors in the model term.

Covariance models

Once a term is specified, a covariance model can be selected for each factor or the whole term, as required:

  • identity
  • autoregressive order 1 or 2
  • moving average order 1 or 2
  • ARMA(1,1)
  • diagonal
  • uniform

See also

VSTRUCTURE directive in command mode for a wider range of covariance models, additional options, and further information about the models. In particular, repeated use of VSTRUCTURE enables covariance models to be specified for multiple model terms.

Updated on April 2, 2019

Was this article helpful?