The responses I have seen seem to have addressed the underlying problem, but did not (I thought) give an explicit answer to the question asked. If I correctly understand your problem, the short answer is "Yes": it is statistically correct to do an analysis of the kind you propose. (Of course, whether you carry it out correctly is another matter!)
A longer answer is embedded in your question, below. (I am not personally familiar with Statistica, so cannot advise you on matters of Statistica syntax, etc.) On Fri, 4 Jul 2003, BarBarella wrote (edited): > Sorry for my English. I need help. I want to do analysis of > variance for repeated measures (two independent variables one of 2 > levels and second 3 levels) and I want to make contrast between > level nr 1 and (level nr 2,3) from the second variable. > Is it statistically correct to do such analysis? This depends in part on your ANOVA results. If the interaction between the two independent variables is negligible, such a contrast, calculated overall (that is, ignoring the two-level factor) is appropriate; but it may be that the pattern of means for the three-level factor might cause you to reconsider that. If the interaction effect is significant, you would want to investigate the pattern of results (that is, of the six cell means) in greater detail, and you almost certainly should examine this contrast (between level 1 and the average of levels 2 and 3) for each level of the two-level factor, separately. > I will make this on Statistica. Thanks for answers. barbarella ----------------------------------------------------------------------- Donald F. Burrill [EMAIL PROTECTED] 56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
