Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

ACBC/HB_Alpha file interpretation - Interactions, Gender and a continuous covariate


I've already read some posts here about this topic with one covariate, but I'm not sure whether I can transfer it to my case.  
I conducted an HB estimation, including 3 interaction effects, gender (1=female, 2=male) as categorical covariate and 'Expect to pay' as zero-centered continuous covariate.  
Ignoring the interaction effects, am I right when interpreting the calculated values of the alpha file shown below as follows:

B for Male=1,49 + 0,33*Expect to pay
B for Female=1,49+0,17+0,33*Expect to pay

Level 1:
Intercept: 1,49
Gender(1): 0,17
Expect to pay: 0,33

My second question is about problems with prohibitions and alternative specific design in the interaction effects.
In my design I set two prohibitions (Attribute2Level1 with Attribute3Level1 and A2L3 with A3L3) and an alternative specific design, in which all Levels of A2 and A3 would not be displayed if A1L1 occured.  Interaction Search led to significant interactions between A1-A3; A2-A3; A1-A2 leading each to an Pct. Cert.-increase of 2%. I included these interactions in my HB. Consequently I get utilities for all combinations, even the ones that never occured together. Can I still interpret the valid combinations and ignore the others or should I forego the interaction effects completely?
Thank you very much for your support in advance!

Best regards
asked Jun 28 by MM (120 points)

Your solution to the original question

Please only use this to answer the original question. Otherwise please use comments.
Your name to display (optional):
Privacy: Your email address will only be used for sending these notifications.
Anti-spam verification:

To avoid this verification in future, please log in or register.