Have an idea?

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

Hierarchical Bayesian Estimation for Choice Based Conjoint Analysis

Dear Everyone,
I am writing my first question on finding ways for Hierarchical Bayesian estimation. I am using Choice Based conjoint studies in B2B marketing scenario where the number of participants is less than 20. The technical papers in Sawtooth library suggest a sample size of 300-500 for any statistical significance for estimating part worth utilities. However, in B2B marketing, I guess this is kinda difficult.
My questions are the following:
1) As I understand from the literature and technical papers, CBC replicates real market place situation where the respondents will be presented with real options. However, owing to the less number of participants, is CBC the correct approach to the task?
2) How do I go about performing Hierarchical Bayesian (HB) estimation to determine the individual part worth utilities for conjoint studies in SPSS or in R? Are there any videos which can be recommended for this? Any advise would be very much appreciated. The reason being, my University is not having Sawtooth license and the only license it has is for IBM SPSS.

Kind Regards,
Sujit Acharya
asked Dec 28, 2017 by sacharya (490 points)

1 Answer

0 votes
The advice about 300-500 respondents would hold for any statistical testing, not just conjoint analysis.   Note that in some markets (B2B, pharmaceutical) it just isn't feasible to get that many respondents and we just have to live with fewer.  

I've had success using CBC with small sample sizes but you might want to consider a traditional ratings-based conjoint (which uses regression analysis as its analytical engine) rather than a choice-based one:  choice-based models use multinomial logistic regression (MNL), which is more demanding in terms of sample size than is regression analysis.

Finally, you can't do hierarchical Bayesian MNL in SPSS but in R you can use the RSG-HB package.
answered Dec 28, 2017 by Keith Chrzan Platinum Sawtooth Software, Inc. (64,975 points)