Abstract: | Contends that tests in repeated-measures designs based on MANOVA are free of sphericity assumptions, and with modern computing software, MANOVA is straightforward to use, even for complex designs and nontraditional hypotheses. A general strategy for implementing MANOVA within statistical computing packages is presented. Regular (preplanned) tests and simultaneous (post hoc) tests are illustrated for a variety of designs and hypotheses. Optimal contrasts for unbalanced repeated measures designs are appended. (33 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved) |