🚀 go-pugleaf

I have two questions:  the first which deals with performing a set
of homoscedasticity tests, and the second which deals with
performing a correlation between an ordinal variable and a
nominal variable.

For the first question:
A client would like to perform a set of comparisons between females
and males for each cell of a 3x3 factorial ANOVA.  Therefore, a
total of 9 pairwise comparisons are requested.  Until I learn that
there are better ways for testing homogeneity of variance for each
cell, I have recommended that PROC TTEST in SAS be run and results
of the homoscedasticity test be used for evaluating the client's
hypotheses.  The null for all these tests is: variances are equal.
The alternative for all these tests is: variance for males > variance
for females.  The set of tests might be conceptualized as follows:

             PA            ST           GENDER
              1             1            males
                                         females    test #1
              1             2            males
                                         females    test #2
              1             3            males
                                         females    test #3
              2             1            males
                                         females    test #4
              ...           ...
              3             3            males
                                         females    test #9

My question is, if it is a valid approach to use a multiple set
of homoscedasticity tests, is there a any kind of adjustment
for the individual comparison error rate to maintain an
'experimentwise' error rate (e.g. alpha=0.05), or can each test
be performed at the alpha=0.05 significance level?

Question 2:
Is there a more appropriate correlation test besides the product-
moment correlation coefficient to use on a set of data when one
of the variables is ordinal (5 point Lichert scale) and one of
the variables is nominal (can take on values 1,2 or 3)?

 --Larry Layne,   Research Consulting Services,  Syracuse University --
 --250 Machinery Hall, Syracuse, NY 13244 USA   (315)443-2196        --
 --Bitnet: LAWORK@SUVM       Internet: LAWORK@SUVM.ACS.SYR.EDU       --