CMStatistics 2018: Start Registration
View Submission - CFE
Title: How to avoid the zero-power-trap in testing for correlation Authors:  David Preinerstorfer - Université libre de Bruxelles (Belgium) [presenting]
Abstract: In testing for autocorrelation of the errors in regression models the power function of, e.g., ``point-optimal-invariant'' or ``locally-best-invariant'' tests can get very low as the correlation gets strong (as sample size stays fixed). We discuss and revisit theoretical results concerning the occurrence of this ``zero-power-trap'' phenomenon, and suggest ways to avoid it.