I am doing a conjoint analysis with stated choice probabilities. I have collected data on smartwatches preferences but asked the respondents not for a single choice but to allocate preferences.

For example, when presented with 3 smartwatches and a no-choice option put a 20%, 30%, 50%, 0%, respectively for each option displayed.

Do I have to use the standard conjoint-package or the mlogit-package or maybe something else?

One strategy would be to break each task into a series of head-to-head comparisons. Create a dataset with one row for each comparison in each task (i.e. watch A vs. watch B in task t) with the difference in each attribute level and the difference in the stated percentages (likelihood of purchase?). Then regress the differences in purchase % against the differences in attribute levels. This would indicate how a 1-unit change in each attribute affects the likelihood of purchase. As a first approximation this could be accomplished with simple linear regression but more advanced models could account for the constrained nature of the dependent variable.

Your Answer

 
discard

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.