1) A optimal and D optimal had the results closely. The author proofed that the maximum value of the element in the item information only depended on the discrimination parameter therefore this result corresponded the expected. The comparison of these methods in item selection was showed in Figure 5. It is reasonable that the items with the higher discrimination parameters at theta1 or theta2 had higher probability to be selected. This figure is a good method to show the exposure rate.
2) it seems that c-composite method can be used in the higher order CAT. After we calibrate the coefficient of second order latent trait and set them as the weight of each latent trait, using c-composite method can select appropriate items for the composited latent trait. But the weight should be treated as random effect for containing the error term of weight.