Dear all,

We have designed an online discrete choice experiment. We have different between-subject experimental groups that differ on how a specific attribute is presented. We want to compare the importance that consumer attach to this attribute according to how it is presented. That is, we want to assess whether the parameter estimate for this attribute is significantly different across the different experimental groups.

Our doubts are the following:

1. What type of model is most appropriate? (MNL, ML, hierarchical Bayesian?)

2. What type of statistical test should we conduct to compare the parameter estimates? (t test, Mann Whitney U test?)

3. Which software should we use to do so? Can we use NLOGIT, STATA or SPSS?

We have been blocked by this for several days now, and would warmly welcome any help.

Kind regards