Re: Help!!! Bob

Bob Green (
Tue, 14 Oct 1997 21:23:36 +1000


This can be a contentious issue, I have been told to have at least 40
participants. See if you can find similar research and how your sample size

With 50 participants, once you have a variable with more than two levels you
are looking at groups of about 17 or less. The power of your analysis will
be limited. However if there is limited work in the area you could justify
your research as exploratory (though I am interested in the views of others).

You could do some form of correlation/regression if you had only one or two
variables. If you examine a multitude of variables then you should use the
Bonferroni criterion. Preferable, is to have some hypothesis that
males/females or some other group will construe differently based on
variables selected from the literature or your experience.

If you want to discuss this further it would be helpful to provide some idea
of the grids you have (size, contents) and other measures (EPQ?). it also
sounds as if you have a few areas of interest;dementia, experiences of
aggression. if you do not have hypotheses, perhaps you should try to
specify your central questions of interest, assuming you haven't already
done this.