In the Manova Fit report, specify the response designs for various tests using the Response Specification panel.
Figure 10.4 Response Specification PanelĀ
Choose Response
Provides choices for response design, which forms the M matrix. The columns of an M matrix define a set of transformation variables for the multivariate analysis. You can choose among the following options for the M matrix:
Repeated Measures
Constructs and runs both Sum and Contrast responses.
Sum
Sum of the responses that gives a single value.
Identity
Uses each separate response, the identity matrix.
Contrast
Compares each response and the first response.
Polynomial
Constructs a matrix of orthogonal polynomials.
Helmert
Compares each response with the combined responses listed below it.
Profile
Compares each response with the following response.
Mean
Compares each response with the mean of the others.
Compound
Creates and runs several response functions that are appropriate if the responses are compounded from two effects.
Custom
Uses any custom M matrix that you enter.
Note: The most typical response designs are Repeated Measures and Identity for multivariate regression. There is little difference in the tests given by the Contrast, Helmert, Profile, and Mean options, since they span the same space. However, the tests and details in the Least Squares means and Parameter Estimates tables for them show correspondingly different highlights. The Repeated Measures and the Compound options show dialogs to specify response effect names. They then fit several response functions without waiting for further user input. Otherwise, selections expand the control panel and give you more opportunities to refine the specification.
Univariate Tests Also
Obtains adjusted and unadjusted univariate repeated measures tests and multivariate tests. Use in repeated measures models.
Test Each Column Separately Also
Obtains univariate ANOVA tests and multivariate tests on each response.
The following buttons are available only after you have chosen a response option:
Run
Performs the analysis and shows the multivariate estimates and tests. See Multivariate Tests in Multivariate Response Models.
Help
Shows the Help for the Response Specification panel.
Orthogonalize
Orthonormalizes the matrix. Orthonormalization is done after the column contrasts (sum to zero) for all response types except Sum.
Delete Last Column
Reduces the dimensionality of the transformation.