Dummy regression for graded paired comparison

#1
Hi all,

I'm trying to create a dummy regression for a conjoint analysis with graded paired comparison.
here is an example of one paired comparison:
Code:
##################              ##################
#     Car 1      #              #     Car 2      #
##################              ##################
#                #              #                #
# Color:   red   #              # Color:   blue  #
# Engine: 100 HP #              # Engine: 110 HP #
# Make:   Honda  #              # Make:   Toyota #
#                #              #                #
##################              ##################

 |-----|-----|-----|-----|-----|-----|-----|-----|
-4                       0                       4
As you can see, the user has to specify his preference for these two fictional car offers. (-4 = strongly prefer Car1; +4 = strongly prefer Car2; 0 = indifferent).
This step will be repeated for multiple paired comparisons.

My problem is, I don't know how to code this into a dummy regression.

In a Conjoint Analysis with a rating or ranking procedure, you would normally use a dummy regression in the form of:

where y is the utility of the actual ranking/rating k and b is the partial utility, to be calculated for every attribute j and attribute's option m.

Is anyone familiar with this concept and can help me creating such a graded paired comparison? Since there exist only proprietary software for such types of conjoint analysis, it's hard to find information about the implementations...