E. Castilla, M. Jaenada Malagón, N. Martín, L. Pardo
While the Morgan-Pitman test is well-known to be the most efficient test for testing uncorrelation for paired normal data, there is not a clear choice for data with mild or moderate outliers. In this work we propose the Wald-type test statistics based on the Rényi's pseudodistance estimators. Theoretically and from an asymptotic point of view, the distribution of the estimators and test-statistic is developed, as well as their influence function analysis. An Iteratively Reweighted Moments Algorithm is provided as a simple way of calculating the estimators. A simulation study as well as real data based example illustrates and support the theoretical results.
Keywords: Robust statistiscal inference
Scheduled
Multivariate Analysis I
June 8, 2022 4:00 PM
Cloister room