I have downloaded a new JMP 11 trial lately to give it a test. There is now an opportunity to directly evaluate consumer's WTP within discrete choice models when you use Consumer research -> Choice.
In the previous version you simply used to calculate substitution rate between a particular factor and price, i.e. dividing model parameter of a factor X by a model parameter of Price. This approach is also supported in literature.
From now, JMP 11 makes it automatically from the individual interface. It provides a change in price which consumer is expecting when moving from value A to value B of a parameter X.
However, what i found is that the calculated WTP is always twice as much as I used to obtain manually in version JMP 10 by simply dividing the relevant parameter by the parameter of price.
Does anybody have an idea, what's the story behind it? Need to verify everything to accomplish my Master Thesis successfully.
Your hand calculations are based on the assumption that your factors are coded 0/1 (dummy coding). JMP uses -1/1 coding by default. So, the WTP calculations are accurate, but they will be different from what you are used to seeing.
If you want JMP to use 0/1 coding, you can change your factors to ordinal instead of nominal.
I am using JMP 15 to perform DCE. If the scenario is selected then it is coded as 1, otherwise it is coded as 0, am I coding it correctly. Does it make a difference in terms of calculating WTP or other outcomes, such as log worth?