I'm doing market (survey) research and have already collected my data. I've already looked at my descriptive statistics and ran numerous correlations. I'm now ready to run regressions in order to find more predictive variables but need help with this. These are my questions:

1: I have a variety of scale, nominal, and ordinal independent variables and am trying to predict my dependent variable, which is "interest in my product"- 5 levels from "not at all" to "very". Given that my DV is not continous and has more than 2 levels, I'm wondering what options I have for regressions. Is my only option ordinal regression?

2: If it is ordinal regression, how do I find the best and most predictive model? Do I simply throw all my potential IV's in and gradually remove the least predictive until I have the best fit? This isn't classic research where I'm trying to prove a hypothesis but instead am simply trying to explore it and find the best.

3: If it is an ordinal regression, do i have to count my ordinal IV's as if they are nominal or can I use them as continuous given that they are 1-5 (very little-A lot) where the order matters? If I can use it either way, is one of the methods best in an ordinal regression?

Thank you!