Date:         Fri, 10 Apr 1998 07:41:24 -0600
Reply-To:     pallier@LSCP.EHESS.FR
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         pallier@LSCP.EHESS.FR
Organization: Deja News - The Leader in Internet Discussion
Subject:      Q: replacement for ANOVA when DV is a small integer?

Hello,
 
I have to analyse the results of an experiment
with 3 factors: 1 between-subjects factor (group)
and two within-subjects factors (S & L).
 
The data presents itself as follow (it's only a short excerpt):
 
Group Subj S1L1 S1L2 S2L1 S2L2
 1     1    5    0    4     0
 1     2    3    0    2     1
 1     3    1    2    0     0
 2     1    1    0    1     2
...
 
These data represent numbers of mistakes (of which there are 4 types)
made in conversation by different groups of people.
 
I would like to compared the counts in the differents
conditions yielded by the three factors.
 
Because the data consists of small integers
with a exponentially decreasing pdf, I believe
that I cannot run a repeated measure ANOVA.
 
I think of running non-parametric tests, that is,
Wilcoxon paired sample tests for the within-subject comparisons
and Mann-Whitney for the between-group comparisons
(within each variable).
 
Is this correct? Are there any other types of analysis
I should consider? I have been told to consider logistic regression
but I do not see how it applies (the data, consisting of counts, belong
to a ratio-scale; it's not categorical, even if it is discrete).
But I maybe be missing something here.
 
Thanks,
 
Christophe Pallier
 
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/   Now offering spam-free web-based newsreading