Power Rates of Simultaneous Tests for Variances

摘要

The report investigates the behavior of power rates associated with simultaneous tests for the variances of a bivariate normal distribution such that separate conclusions can be drawn about each parameter. The power rate is defined as the probability that one or both hypotheses are rejected when either one, or both, are not true. Alternate hypotheses were investigated when both hypotheses have either (i) two-sided, (ii) one-sided upper, or (iii) one-sided lower alternatives. (Author)