I have two questions:
1. For Figure 1, it is similar to the higher-order model. However, in the current study, it was assumed that the thetas in three nodes follow the multivariate normal distribution.What's the rational for the assumption?
2. Can the model in Figure 1 be reduced to two levels, with the first level be the four categories of response and the second level be the g theta?