I am working on a dataset, which involves three categorical (two
dichotomies, and one of 5 categories) independent variables and a continuous
variable that consists of counts. The idea is to look at the significance of
the group differences/ the (main and interaction) effects of the
independents on the dependent. The problem is that a) some groups formed by
the independents are not at all equal in sample size (1 group has no people
left inside), and b) the dependent variable is not normally distributed and
has extreme outliers, which as part of the project should not be deleted. So
far, I have tried factorial ANOVA (though violating most of its
assumptions), Hierarchical LogLinear analysis, and multiple regression with
the independents entered as dummies. However, none of the above tests seem
appropriate. Any suggestions? Thank you ~ Anna
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
