What Thom posted covers the statistical part - Here are a couple of minor considerations.
On Thu, 11 Dec 2003 13:36:47 +0800, "Erica So" <[EMAIL PROTECTED]> wrote: > I have one more que about the Repeated Measure Anova. > > "Dot' appears in some of the output of the Repeated Measure, e.g. all output > of the Multivariate Tests; the Chi-aq, Sig and Huyuh-Feldt (except the > interaction term) of the Mauchly's Tests of Sphericity; SS, df, MS, F and > Sig of Huynh-Feldt in the Tests of Within-subjects Effects; F and Sig of the > Tests of Between-subjects Effects. What do those dots mean? And what are > the possible causes of the appearance of 'Dot'? For SPSS, the 'dot' is used by case listings as the 'system missing' indicator: no data provided. I assume that you are seeing the same code, essentially, in the output from SPSS or whatever package (unnamed). An analysis was not produced, because it was inappropriate or the data were not sufficient. > > In addition, 'cannot produces multivariate test stat because of insufficient > residual degrees of freedom' was appeared at the bottom of the Multivariate > Tests. As Thom says, some tests (sphericity) are not defined for just 2 periods, and so on. However, this error/ warning sounds like a less common one, and it prompts me to ask: - Did you have any missing data within cases? - Are you aware that most MV programs will *drop* all the data in a record, if any single score for the analysis is missing? On the one hand, you can look at the univariate descriptions (means, SDs, Ns) to confirm that there is no Missing. On the other hand, you can check what the analysis says that it included, in order to see that it has all the data -- and make sure that you did not accidentally drop groups, or time-periods, etc., owing to mis-specification. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html "Taxes are the price we pay for civilization." . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
