On 19 Mar 2002 14:33:17 -0800, [EMAIL PROTECTED] (George Caine) wrote: > I am completing my results section for my dissertaion and have decided > to use a 2by3 analysis of variance. I know this is the most effective > test statistic to apply to my study. However, I am worried that my > result section may be too short and feel it doesnt allow me to > demonstrate as much statistical knowledge as I would like. Is there > any subsequent analysis I could apply or does it matter?? any advice > would be greatly appreciated.
Don't you have to do something to validate your analysis? Numerical work usually collects things like age and sex, and date (serial sequence number), and so on -- which comprise factors that are, you might presume, irrelevant to outcome. Part of validation is (say) showing that the measure you have *does* measure something, and another part is (say) showing that it does not relate to things that it is supposed to not-relate to. - On the one hand, if the quantities you have make up the 2x3 table, and that is enough for your dissertation, then your field doesn't seem to demand statistical knowledge. So your concern might be misplaced. - On the other hand, you might show that the distribution (at least) is apparently normal with similar variances in the groups. Hope this helps. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
