- Both CVM and CM are based on RUM. If one is theoretically invalid the other is as well.
- Both CVM and CM are subject to GIGO. Unfortunately, there was a lot of garbage in, in some of the early CVM work, which was implemented by people without much experience in survey research. Since CM developed a little later, these researchers could avoid the mistakes of the 1970s and early 1980s. The results are only as good as the experimental design and the survey.
- I generally use CM because I am usually in a multi-attribute framework and I think CM is more useful for this.
- For one attribute (or a set of attributes that vary in lockstep) CVM and CM are conceptually equal.
Jim is using CM as a acronym for choice modelling, another name for choice experiment.
I agree with each of these points but REALLY like point #2 and especially agree that #3 is an excellent reason to choose choice experiments.
Regarding #4, notice that the title of this post contains "and" and not "vs".