back transformation in a Mixed model ANOVA<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />I apply a log transformation before an ANOVA - split-plot mixed model. The least squares means in the output correspond to a back-transformation = EXP(Y) wich is the median. The arithmetic mean is EXP(Y + variance/2). In a split-plot, there is more than one variance. Which variance should I use ?
Thanks.
