On a nonparametric test for the mixed-model

From MaRDI portal





Nonparametric tests for the null hypothesis of no treatment effect in the mixed-model experiment which involves n randomly chosen subjects who respond once to each of p distinct treatments have been developed by \textit{G. G. Koch} and the second author [Biometrics 24, 27-48 (1968)]. These tests were based on the assumption of compound symmetry of the error vectors and on the weaker assumption of diagonal symmetry of the error vectors. This paper considers an alternative (permutationally) distribution-free test under this latter assumption. The new test follows the same type of distribution theory as in the cited paper, but utilizes the inherent invariance structure in a more visible and direct way.











This page was built for publication: On a nonparametric test for the mixed-model

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3805650)