METHOD Subcommand (VARCOMP command)

The METHOD subcommand offers four different methods for estimating the variances of the random effects. If more than one METHOD subcommand is specified, only the last subcommand is in effect. If the subcommand is not specified, the default method MINQUE(1) is used. METHOD cannot be specified without a keyword.

MINQUE(n). Minimum norm quadratic unbiased estimator. This method is the default method. When n = 0, zero weight is assigned to the random effects and unit weight is assigned to the residual term. When n = 1, unit weight is assigned to both the random effects and the residual term. By default, n = 1.

ML. Maximum likelihood method. Parameters of the fixed effects and variances of the random effects are estimated simultaneously. However, only the variances are reported.

REML. Restricted maximum likelihood method. Variances of the random effects are estimated based on residuals of the model after adjusting for the fixed effects.

SSTYPE(n). ANOVA method. The ANOVA method equates the expected mean squares of the random effects to their observed mean squares. Their variances are then estimated by solving a system of linear equations. The expected mean squares are computed based on the chosen type of sum of squares. Two types are available in VARCOMP: Type I (n = 1) and Type III (n = 3). Type III is the default option for this method.