Have an idea?

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

Design (# of choice sets) Help

I am new to the method.

I'm considering a CBC with a 2 x 2 x 2 x 3 design (no prohibitions, no alternative specific constants).

How do I determine how many choice sets a single participant is exposed to?
(I want to minimize the number, which I think means I'd want to show more than 2 options and at time and maybe need a high number of total participants. I just don't know how to play around/think about with these inputs).
asked Apr 20 by anonymous

1 Answer

0 votes
Fortunately, you have a very small attribute list with just 4 attributes and 9 total levels.  The number of choice tasks to ask each respondent is a common question.  Usually, CBC practitioners choose between 8 and 14 choice tasks, which take a typical respondent somewhere around two to three minutes.  

Most CBC practitioners want to estimate stable individual-level utilities using HB analysis.  This leads to more accuracy and flexibility for market predictions and market segmentation than if you ask fewer choice tasks per respondent and are limited to aggregation techniques like aggregate logit.

A rule of thumb we teach in our trainings is that for reasonable HB estimation in CBC, you want each level to be shown to each respondent six or more times.  With only 2 or 3 levels per each attribute, this is very easily attainable if you show three product concepts (alternatives) per task and show even as few as six choice tasks.   

If your attribute levels have complex (long) text, then things may not be as easy as it appears for your study.  So, you'll have to consider that.  

Assuming your attribute levels are reasonably compact, given your small attribute list, I'd think 8 choice tasks each with 3 product concepts should be sufficient.
answered Apr 20 by Bryan Orme Platinum Sawtooth Software, Inc. (186,465 points)
Thank you!