Overview (VARCOMP command)

The VARCOMP procedure estimates variance components for mixed models. Following the general linear model approach, VARCOMP uses indicator variable coding to construct a design matrix and then uses one of the four available methods to estimate the contribution of each random effect to the variance of the dependent variable.

Options

Regression Weights. You can use the REGWGT subcommand to specify regression weights for the model.

Estimation Methods. You can use the METHOD subcommand to use one of the four methods that are available for estimating variance components.

Tuning the Algorithm. You can control the values of algorithm-tuning parameters with the CRITERIA subcommand.

Optional Output. You can request additional output using the PRINT subcommand.

Saving the Results. You can save the variance component estimates and their asymptotic covariance matrix (if produced) to an external data file.

Basic Specification

The basic specification is one dependent variable and one or more factor variables (that define the crosstabulation) and one or more factor variables on the RANDOM subcommand (to classify factors into either fixed or random factors). By default, VARCOMP uses the minimum norm quadratic unbiased estimator with unit prior weights to estimate variance components. Default output includes a factor-level information table and a variance component estimates table.

Subcommand Order

• The variable specification must come first.
• Other subcommands can be specified in any order.

Syntax Rules

• Only one dependent variable can be specified.
• At least one factor must be specified after BY.
• At least one factor must be specified on the RANDOM subcommand.