I am conducting a simple CBC study, for brand preference. The client has asked that we show some brands all of the time, and others some of the time, to reflect the existing distribution in stores. I am interested in knowing the best way to do this.
I have done this by having one master attribute for brand, and then alternative specific attributes for the brands that are shown selectively. Two I have used binary attributes for "on" / "off", and one I have used a 3 level attribute to reflect it being available a 1/3 of the time.
Is this the best way to achieve this type of design, and are there any pitfalls I should watch out for? Currently the design appears to be well specified, with standard errors below 0.5 for main effects.
In addition, as a separate question, the client has asked that we do not vary prices within each task, but instead impose variation across tasks. We will not be measuring price-preference, and could measure brand preference using simple counts. In theory we could look at preference vs. none as all prices shift up and down in different tasks. Has anyone done anything like this before?
Thank you in advance