Have an idea?

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

Running a conjoint across countries with varying attributes

I need to run a choice-based conjoint in several different countries. I will have the same features in all of the countries, however two of my attributes will need to have different levels for each different country (brand and price).

My client wants an optimized version of the product for each of the countries. Considering that prices and brands won’t be standardized do you have any source or technical paper that could enlighten me about the potential risks or safe steps to take for this conjoint to be successful?
asked Feb 23, 2017 by JKincaid Bronze (1,035 points)

1 Answer

+1 vote
I don't have a particular reference to offer but I can imagine a couple of ways of doing this.  

I would probably estimate utilities for each country separately, assuming you have enough sample size to do so.  If not, and this will be more difficult, I would field the survey as if each country were separate then bring the countries into a single data file where I coded the brand and price variables to be country-specific.  I'd probably add county as a covariate and then run the analysis on this larger data matrix.  This will allow more stability for the  utility estimates of your non-price, non-brand attributes, assuming preferences for them don't differ a lot by country.  

Like I say, this will be more work because you have to combine all the country's data files into one and then code the country-specific brand and price attributes; then you pull that larger design into CBC/HB to analyze.    I would probably analyze this as a separate model per country if my sample size allowed.
answered Feb 23, 2017 by Keith Chrzan Platinum Sawtooth Software, Inc. (103,325 points)