Statistical test over binary data

#1
Given certain products (ProductA, ProductB, ProductC, etc..) users are asked to rank them according to theirpreference (whether they like , somewhat like or not like the product).

Thereafter I have extracted certain binary features from these products (handle is present or not, etc..) and now I want to test which features are prominent (statistically significant) in user liked, somewhat liked and not liked products.

What sort of a statistical test I should use for this sort of a problem and why?

Any help/advice would be really appreciated.
 
Last edited:

BGM

TS Contributor
#2
You may want to fit an ordered logistic regression model

http://en.wikipedia.org/wiki/Ordered_logistic_regression

You have an ordinal dependent variables with 3 levels;

You have some binary independent variables as well and you want to study their relationship.

So you may try to test whether the corresponding coefficients is significant.

I am not familiar with that technique so correct me if I am wrong.