Answers are grounded in this paper and cite their sources.
Computer sampling from a multivariate normal spherical population was used to evaluate Type I error rates for a test of P = I based on Fisher’s tanh −1 variance stabilizing transformation of the correlation coefficient. The range of variates considered was 5 to 25 and Type I error rates were estimated for several sample sizes with 2,500 independent replications. Except for small samples the test was well behaved. After the test converges to an acceptable Type I error rate it is preferable to Box’s test of P = I.
No takes yet. Share an insight, caveat, or question.
John R. Reddon (1987) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: