SUMMARY Outlier discordancy tests are motivated by considering the residual pattern recognition approach of Daniel and the approaches of cross-validation and maximum likelihood. Various optimality properties of the significance tests for the most outlying subset of data are shown and a multivariate two-way layout example is given.
Nurisha Mohd ZulkefliAdzhar RambliMohamad Ismeth Khan Azhar SuhaimiIbrahim MohamedRaiha Shazween Redzuan
Wobbe P. ZijlstraL. Andries van der ArkKlaas Sijtsma
Surendra P. VermaLorena Díaz‐GonzálezMauricio Rosales-RiveraAlfredo Quiroz‐Ruiz
Bruno GiraudeauClaude Chastang