Hello everyone!
I have a few questions related to max-diff questionnaire design when we consider taking two-dimensional data on the same choice sets(in my case importance and satisfaction data):
My study is to prioritize attributes based on the importance and satisfaction data on water transport-related attributes by using max-diff scores in Importance-Performance Analysis Matrix. I have 20 attributes in total and I am thinking to take 5 attributes per choice set and limit the number of questions per respondent as much as possible because of the cognitive burden. As I want to collect importance and satisfaction data with respect to the same sets, if I repeat each attribute 3 times per respondent, then 12 choice sets should be answered for two dimensions which is practically impossible.
1. Will it be fine if I take 8 or less choice sets and still get good estimates? In my case, how many attributes per choice set would be better (4 or 5)? For paper-pencil interviews, how many versions are to be considered for better results? Which analysis can be used to get max-diff scores? What is the suggested sample size?
2. If Balanced Incomplete Block Design (BIBD) is used which is specified in most of the research articles for creating best-worst scaling, it is said that there are only a limited number of possible combinations of attributes forming choice sets i.e., BIBDs. So, the total number of choice sets to be prepared in total for efficient design is decided. Once choice sets are prepared, they are blocked into say 5 blocks(versions) each with some choice sets that a respondent could answer easily. But, in Sawtooth software we can't say how many choice sets in total are to prepared. We discuss versions here. How can anyone decide the total number of choice sets to be prepared in total (including all versions) for a particular total number of attributes and the number of attributes per choice set for an efficient design?
Can anyone please clarify my confusion and how to justify my work without using BIBD?
Thanks!
Vaishnavi.