Have an idea?

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

Technical paper on the use of binary attributes in a CBC analysis

in my CBC analysis I want to find out what kind of impact Corporate Social Responsibility (CSR) has on a buyer's decision to choose an airline. I have five different attributes (Price, Seating Comfort, In-Flight Meals, Departure Time, and CSR). All attributes, except for CSR, will have three levels. Since I want to find out which CSR dimension (economic, social, ecological) has the biggest impact I would have 8 different levels if I want them to be mutually exclusive (all, none, econ., social, ecol., econ. & ecol., econ. & social, social & ecol.).
To avoid a number-of-levels effect I thought it would be best to transform this 8-level-attribute into three binary attributes (Economic Sustainability, Social Responsibility, and Ecological Responsibility).
Unfortunately I can only find technical papers regarding the number-of-levels effect but non that would strengthen my decision to use three binary attributes. My supervisor is worried that three  binary attributes regarding CSR will also have a negative effect on the respondents because three out of seven attributes will be concerned with CSR.
It would really help to find a technical paper that confirms my decision. Any help would be highly appreciated.
Thank you very much.

Best regards
asked Dec 24, 2017 by Carla Merz (230 points)

1 Answer

0 votes

I don't see a big problem here one way or the other.  Also, even though you can choose to have an 8-level variable or split it out to three binarys, to the respondent those two things will look the same, right?   Respondent sees the stimuli, not how you've coded up your experimental design.
answered Dec 24, 2017 by Keith Chrzan Platinum Sawtooth Software, Inc. (62,700 points)