This time, the identified component

This time, the identified components do not appear to repre- sent either high or low scorers. There is a noticeable contrast in the liking scores between bread B, which was a very seedy bread. Bread N, a hearth bread, is also clearly polarizing. There are also some breads that neither group of panelists particularly liked, e.g., bread D, and others that both groups liked similarly, e.g., bread C. Perhaps the most convincing argument, however, that these results do not reflect usage of the scale is that the average scores for the two clusters are very close (5.75 vs. 5.56).
Again, we consider nested 16 P3 and 16 P4 SIDs. The CUU model was fitted to each nested design for G 1⁄4 1; . . . ; 4 components and q 1⁄4 1; . . . ; 6 latent factors. In both cases, the best fitting CUU mod- els have G 1⁄4 2 components and q 1⁄4 1 latent factor, with BIC values of