I was asked to look at some data that was generated using a best-worst scaling experiment (or at least, that is what was intended). Having some knowledge of BWS I felt confident that I could help. However, there is an aspect of the design that I cannot get my head around and I cannot find any literature or online resources for guidance. So, I am hoping somebody might be able to point me in the right direction.
Essentially, the design contains 21 choices. So it looks like case 1 BWS. Each respondent received 7 choice sets with 3 choices per set and they selected the best and worst option within each set. Here is the part I do not understand. The choice sets were divided into 7 blocks and respondents were randomly assigned to one block (each block contained 7 choice sets). Therefore, respondents were not exposed to the same choice sets, although they did see all 21 items. My questions is, how do you analyse choice preferences from this type of design given that the choice sets respondents received were heterogeneous? Any guidance would be greatly appreciated.