Overview (GLM: Multivariate command)

This section discusses the subcommands that are used in multivariate general linear models and covariance designs with several interrelated dependent variables. The discussion focuses on subcommands and keywords that do not apply—or apply in different manners—to univariate analyses. The discussion does not contain information about all subcommands that you will need to specify the design. For subcommands that are not covered here, see GLM: Univariate.

Options

Optional Output. In addition to the output that is described in GLM: Univariate, you can have both multivariate and univariate F tests. Using the PRINT subcommand, you can request the hypothesis and error sums-of-squares and cross-product matrices for each effect in the design, the transformation coefficient table (M matrix), Box’s M test for equality of covariance matrices, and Bartlett’s test of sphericity.

Basic Specification

• The basic specification is a variable list identifying the dependent variables, with the factors (if any) named after BY and the covariates (if any) named after WITH.
• By default, GLM uses a model that includes the intercept term, the covariates (if any), and the full factorial model, which includes all main effects and all possible interactions among factors. The intercept term is excluded if it is excluded in the model by specifying EXCLUDE on the INTERCEPT subcommand. GLM produces multivariate and univariate F tests for each effect in the model. GLM also calculates the power for each test, based on the default alpha value.

Subcommand Order

• The variable list must be specified first.
• Subcommands can be used in any order.

Syntax Rules

• The syntax rules that apply to univariate analysis also apply to multivariate analysis.
• If you enter one of the multivariate specifications in a univariate analysis, GLM ignores it.

Limitations

• Any number of factors can be specified, but if the number of between-subjects factors plus the number of split variables exceeds 18, the Descriptive Statistics table is not printed even when you request it.
• Memory requirements depend primarily on the number of cells in the design. For the default full factorial model, this equals the product of the number of levels or categories in each factor.