Specifically,
My response is a 0, 1 or >=2 ordinal variable
My predictors are 1 categorical ( three categories) and 3 count variables ( each with 22,41,and 36 zeros)
It is usual practice to downgrade higher order scale data to a lower order in order to be able to do some types of analyses, but it is not the practice to upgrade a lower order scale data to a higher order scale data. Specifically, I doubt the validity of upgrading categorical scale data to ordinal data for the purpose of modelling with
ordinal variables.
You can find your answer here
http://support.sas.com/resources/papers/proceedings12/427-2012.pdf
http://support.sas.com/resources/papers/proceedings12/427-2012.pdf
Thank you for your responses, much appreciated. However, for ordinal response, two most common techniques that I am aware of are:
Ordered logistic regression OR Ordered probit regression.
One could (although it is not recommended) use Multinomial logistic regression, which is similar to doing ordered logistic regression, except that it is assumed that there is no order to the categories of the outcome variable (i.e., the categories are nominal, which is NOT my case).
Anyways, that was not my question, my question is if there are any techniques developed to incorporate what happens when in addition to ordinal response the predictors are zero inflated count.
MicrOsiris has MNA (multivariate nominal scaled analysis. Description can be found in the MicrOsiris User Guide at microsiris.com.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS