Since you have a 2-groups problem, Discriminant
computes 1 discriminant function. In the Wilks'
Lambda table, the right-most entry is labeled
"Sig." Do not interpret the value of Wilks' lambda,
instead look at the value of Sig. If it is a relatively
small value, say less than 0.05 by convention, then
you have a statistically significant difference in
means by groups, and therefore have discriminability.
In this instance, then you can look at other parts
of the output to determine which variables are
contributing to discrimination.
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of
Robert Marshall
Sent: Wednesday, July 19, 2006 7:14 AM
To:
[hidden email]
Subject: Wilk's lamda interpretation question
Question that needs more statistical experience than have. Here it is:
Results of discriminant analyis were Wilk's lamda of .943, which I interpret
to mean that the two groups were NOT significantly different, statistically.
However, the resulting structure matrix does identify one variable related
to the function identified. Intuitively, the variable is important in
discriminating between groups, as the structure matrix correctly identified,
even though the Wilk's lamda is very high which suggests no significant
differences between groups. QUESTION: Is it OK to say that the variable is
IMPORTANT," in discriminating between the two groups, even though the
analysis shows NO significant difference between groups? Hope this makes
sense.
Robert
Locked
