Hello all.
I am slightly confused about the McNemar test. I have binary data generated
before and after experimental intervention. My understanding is that the
McNemar test will allow me to determine whether my intervention had a
significant effect. Siegel's book on nonparametric stats (1988) explains
the rational behind the test perfectly well and includes the formula, which
is adopted from chi-square.
The problem arises when you get SPSS to carry out the test; it merely gives
you a probability value. Upon asking a colleague he suggested that it was
possible that the test SHOULD only produce a probability value. When I
sought 'clarification' through the web in general, I found various
explanations of the McNemar test. For example
http://www.itl.nist.gov/iaui/894.01/sigtests/mcnemar.htm suggests that the
McNemar test is merely a method of determining the probability of finding
two sets of binomial results which are inconsistent.
Can anyone offer me any clarification on what seem to be opposing
interpretations?
Thanks in advance for your time,
Michael Gormley.
---------------------------------
Michael Gormley
Department of Psychology
University of Dublin
Trinity College
Dublin 2
email: [log in to unmask]
---------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|