On 10/16/06 1:09 PM, Alida <
[hidden email]> wrote:
> I'm running a GLM to test a MANOVA (I'm doing my masters thesis) and I'm
> confused by the results of the Mauchly sphericity test.
>
> Is the sphericity assumption violated or assumed if:
> Mauchly W = 1; df=0; and epsilon =1?
>
> Or do I simply have to use a different test in my 2x2 factorial design (e.g.
> Levene)?
>
> I could'nt find any aquivalent results in the literature! Do you know any
> reference?
Mauchly's test is assessing the condition of sphericity, which is when the
variance of differences between levels of repeated measures factors are
equal. It is a way of addressing assumptions about the structure of the
covariance matrix in repeated measures designs.
If all your repeated measures factors have only two levels, there is only
one covariance for each repeated measure. Therefore, the assumption about
the covariance matrix is always met and a test of sphericity is not
relevant. Unfortunately, the GLM command of SPSS appears to print the test
of sphericity even under these conditions, which creates the confusion you
seem to be having.
-Jeffrey Berman
Locked
