PRINT Subcommand (GLM: Multivariate command)

By default, if no PRINT subcommand is specified, multivariate GLM produces multivariate tests (MANOVA) and univariate tests (ANOVA) for all effects in the model. All PRINT specifications that are described in PRINT Subcommand (GLM: Univariate command) are available in multivariate analyses. The following additional output can be requested:

TEST(SSCP). Sums-of-squares and cross-product matrices. Hypothesis (HSSCP) and error (ESSCP) sums-of-squares and cross-product matrices for each effect in the design are displayed. Each between-subjects effect has a different HSSCP matrix, but there is a single ESSCP matrix for all between-subjects effects. For a repeated measures design, each within-subjects effect has an HSSCP matrix and an ESSCP matrix. If there are no within-subjects effects, the ESSCP matrix for the between-subjects effects is the same as the RSSCP matrix.

TEST(MMATRIX). Set of transformation coefficients (M) matrices. Any M matrices that are generated by the MMATRIX subcommand are displayed. If no M matrix is specified on the MMATRIX subcommand, this specification is skipped, unless you are using a repeated measures design. In a repeated measures design, this set always includes the M matrix that is determined by the WSFACTOR subcommand. The specification TEST(TRANSFORM) is equivalent to TEST(MMATRIX).

HOMOGENEITY. Tests of homogeneity of variance. In addition to Levene’s test for equality of variances for each dependent variable, the display includes Box’s M test of homogeneity of the covariance matrices of the dependent variables across all level combinations of the between-subjects factors.

RSSCP. Sums-of-squares and cross-products of residuals. Three matrices are displayed:

  • Residual SSCP matrix. This matrix is a square matrix of sums of squares and cross- products of residuals. The dimension of this matrix is the same as the number of dependent variables in the model.
  • Residual covariance matrix. This matrix is the residual SSCP matrix divided by the degrees of freedom of the residual.
  • Residual correlation matrix. This matrix is the standardized form of the residual covariance matrix.

Example

GLM Y1 Y2 Y3 BY A B
 /PRINT = HOMOGENEITY RSSCP
 /DESIGN.
  • Since there are three dependent variables, this model is a multivariate model.
  • The keyword RSSCP produces three matrices of sums of squares and cross-products of residuals. The output also contains the result of Bartlett’s test of the sphericity of the residual covariance matrix.
  • In addition to the Levene test for each dependent variable, the keyword HOMOGENEITY produces the result of Box’s M test of homogeneity in the multivariate model.