Test procedures in coyariamce analysis assuming a general correlation pattern

From MaRDI portal





The analysis of covariance procedure is considered when the observations in each cell are equicorrelated. A correction procedure is given. A computationally easier conservative test statistic is also given. The conservative test statistic allows one to more readily determine the consequences of ignoring correlations, even slight correlations, in the analysis of covariance procedure. The difference of the corrected test and the conservative test is shown to converge in probability to zero. This conservative test is easy to implement on statistical computer packages. It is shown, that for the general correlation pattern, any test involving the regression coefficients of the covariables is an exact test. An example illustrates the procedure.











This page was built for publication: Test procedures in coyariamce analysis assuming a general correlation pattern

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3747545)