Darren is of course correct, but I hope the following, brief, intentionally non-technical explanation will help: Repeated measures analyses are needed only when there are three or more measurements from a given experimental unit and the two or more measurements are both on the right hand side of the equals sign, i.e. are independent variables. When there are only two observations, any one of the following models can be used, none of which have two measurements from the same experimental unit as independent variables.: change (i.e. post-pre)=pre post=pre You will note that in both models only one observation from a given subject is on the right hand side of the equals sign, i.e. pre. When you have three observations from a given subject you generally need to have two or more observations on the right and side of the equals sign (unless you are doing something like a Markov Chain, but Markov Chains are beyond the scope of this Email.) and so need to consider repeated measures techniques. John John Sorkin M.D., Ph.D. Chief, Biostatistics and Informatics Baltimore VA Medical Center GRECC and University of Maryland School of Medicine Claude Pepper OAIC University of Maryland School of Medicine Division of Gerontology Baltimore VA Medical Center 10 North Greene Street GRECC (BT/18/GR) Baltimore, MD 21201-1524 410-605-7119 - NOTE NEW EMAIL ADDRESS: [EMAIL PROTECTED]
>>> Darren Weber <[EMAIL PROTECTED]> 5/11/2005 6:40:09 PM >>> With respect to calculating the epsilon index of sphericity for ANOVA, discussed on pp. 45-47 of: http://www.psych.upenn.edu/~baron/rpsych.pdf It notes that epsilon is not required for a repeated measures design with only k=2 levels, as the minimum value of epsilon (e) is given by: e = 1/(k-1) so for k=2, we have e = 1 (ie, no correction of the F test df; see p. 46). These notes apply to a univariate F test. How do we estimate the minimum value of epsilon for a 2 factor ANOVA? I have an experiment where we measure brain activity from the left and right hemisphere, for two experimental conditions, in each subject. I consider the measures from each hemisphere a repeated measures factor (2 levels) and the experimental conditions is also a repeated measure (2 levels). The question now is, how do we calculate epsilon for this 2 factor study and is it possible that epsilon could be anything < 1 when each factor has only 2 levels? [[alternative HTML version deleted]] ______________________________________________ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html [[alternative HTML version deleted]] ______________________________________________ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
