Determining the Most Significant Parametric Function for a Given Linear Multivariate Hypothesis

Abstract

After a hypothesis about some linear multivariate statistical model has been tested and rejected (e.g., in a MANOVA), many researchers employ simultaneous test procedures to locate the source(s) of the rejection. If the global test was conducted using Roy’s largest root criterion, then this procedure guarantees at least one linear combination of the model parameters relative to some linear combination of the dependent variables that is significantly different from its hypothesized value. This most significant parametric function is not always easy to find, however, because it may not manifest itself in simple or “obvious” functions. A general solution to this problem is presented along with a practical example of its application.