The aim of the paper is to estimate some elements of Q-matrix in Bayesian inference. Firstly, we will focus some uncertain elements by subjective judgment. Then those probabilities can be estimated in Bayesian way. It was easily implemented in OpenBUGS. Many conditions were manipulated including the element choose, misspecified elements, and the number of uncertainty in Q-matrix.
1. The target q elements were chosen by the experimenter himself in advance. Then conducting the Bayesian inference. In fact, this is sort of relying on expert knowledge and confirmatory way. It may count on exploratory way as well in future study.
2. How many uncertain q elements are allowed to obtain satisfactory estimation? How many sample size of people or number of item?
3. Condition 4 showed very good results when all skills were specified as random variable and estimated. One way to improve the results is to include the expert judgment (i.e., prior information). Why it performed worst in item 1 should be examined.
4. DIF study may be possible.