14.03.2014 Views

Modeling and Multivariate Methods - SAS

Modeling and Multivariate Methods - SAS

Modeling and Multivariate Methods - SAS

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Chapter 5 Fitting Multiple Response Models 161<br />

Response Specification<br />

Table 5.4 Descriptions of the Choose Response Options<br />

Repeated Measures<br />

Sum<br />

Identity<br />

Contrast<br />

Polynomial<br />

Helmert<br />

Profile<br />

Mean<br />

Compound<br />

Custom<br />

Constructs <strong>and</strong> runs both Sum <strong>and</strong> Contrast responses.<br />

Sum of the responses that gives a single value.<br />

Uses each separate response, the identity matrix.<br />

Compares each response <strong>and</strong> the first response.<br />

Constructs a matrix of orthogonal polynomials.<br />

Compares each response with the combined responses listed below it.<br />

Compares each response with the following response.<br />

Compares each response with the mean of the others.<br />

Creates <strong>and</strong> runs several response functions that are appropriate if the<br />

responses are compounded from two effects.<br />

Uses any custom M matrix that you enter.<br />

The most typical response designs are Repeated Measures <strong>and</strong> Identity for multivariate regression. There<br />

is little difference in the tests given by the Contrast, Helmert, Profile, <strong>and</strong> Mean options, since they span<br />

the same space. However, the tests <strong>and</strong> details in the Least Squares means <strong>and</strong> Parameter Estimates tables for<br />

them show correspondingly different highlights.<br />

The Repeated Measures <strong>and</strong> the Compound options display dialogs to specify response effect names.<br />

They then fit several response functions without waiting for further user input. Otherwise, selections<br />

exp<strong>and</strong> the control panel <strong>and</strong> give you more opportunities to refine the specification.<br />

Custom Test Option<br />

Set up custom tests of effect levels using the Custom Test option.<br />

Note: For instructions on how to create custom tests, see “Custom Test” on page 68 in the “Fitting<br />

St<strong>and</strong>ard Least Squares Models” chapter.<br />

The menu icon beside each effect name gives you the comm<strong>and</strong>s shown here, to request additional<br />

information about the multivariate fit:<br />

Table 5.5 Description of the Custom Test Options<br />

Test Details Displays the eigenvalues <strong>and</strong> eigenvectors of the E – 1 H matrix used to<br />

construct multivariate test statistics.<br />

Centroid Plot<br />

Plots the centroids (multivariate least squares means) on the first two<br />

canonical variables formed from the test space.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!