The data I'm working with contains more than 50 groups available for multiple group analysis. In your Topic 1 handout you state that multiple group analysis is difficult to carry out with many groups (p. 203). I'm considering doing a cluster analysis to reduce the number of groups I'm dealing with and I have been searching for some guidelines on what constitutes "many" groups.
Do you have any references you could suggest that would clarify this?
More than 10 groups is many. I would suggest creating 49 dummy variables to see where important group differences may be found in the intercepts. Regress the factor on the dummy variables. Include MODINDICES (ALL); in the OUTPUT command. See where the regression of the y's on the dummy variables show intercept differences.