logistic regression?

#1
I am having some difficulty deciding on the correct test to use in a project I'm working on. I have one DV with four categories. I have three IV's with 3, 3, and 2 categories each. I need to do a regression procedure to determine which of the IV's best predict the DV in each category. I think I need to do a logistic regression, but frankly, I'm a novice at this and am out of my element. Anyone willing to help? Many thanks. :confused:
 
#4
Hi Traveler

I regularly have to model the type of problem you are describing. You could use logistic regression(polychotomous), descriminant analysis or decision trees(CHAID). I would not use ANOVA because it requires that your DV is interval data (ie real valued).
 
#6
what about chi-square?

Would chi-square be better? My DV is Anxiety level (minimum, mild, moderate, severe). The IV's are prior history of anxiety (yes/no), percentage of loss of possessions (0-33%, 34-66%, 67-100%), and time evacuated from home (three time period categories).