Carmen, The highlighted numbers are p-values for testing the hypothesis
that the means in the same subset (column) directly above are equal (in
the statistical sense).
The Duncan post-hoc test that SPSS uses is based on Duncan's multiple
range proceedure and is explained in most good middle level statistical
texts.
The SPSS Duncan tables are interpreted as follows.
The means are sorted from smallest (top) to largest (bottom). Means
appearing in the same subset (column) are not considered significantly
different. If a mean appears in only one column then it is significantly
different from all means not in its column. If a mean appears in
multiple columns then it is not considered sig. different form any
means in the columns it appears in. This latter situation occurs when a
mean is somewhere between means that are sig. different (i.e. between
two extremes).
Note that Post-hoc tests like Duncan's should only be used after an
ANOVA has been carried out and the hypothesis of equal means rejected.
You can think of it this way. If the ANOVA leads to rejection of the
equal means hypothesis you then have your licence to go hunting for
which pairs of means might be different.
--
Dr Graeme Byrne
La Trobe University, Bendigo
PO Box 199, Bendigo, 3552
Phone: 61 3 5444 7263
Fax: 61 3 5444 7998
[EMAIL PROTECTED]
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
