Menentukan ukuran matriks Q pada model DINA untuk dijadikan dasar menyusun item tes diagnostik

This study measures an efficient Q-matrix value using Monte Carlo simulation. The simulated Q-matrix, obtained from the simulation conducted based on the real data set of the responses of the seventh graders of middle schools, has 3, 4, and 5 attributes. The middle schools include SMPN 1 Panjatan Kulon Progo, SMPN 1 Sanden Bantul, and SMPN 1 Yogyakarta. The information criteria value used in determining the efficient Q-matrix are the Akaike value (AEC) and the Bayesian (BIC). The simulation results in Q4×3 matrix for the 3 attributes, Q6×4 matrix for the 4 attributes, and Q6×5 for the 5 attributes. Those matrixes are recommended in developing diagnostic test items within DINA model framework. In other words, in the development of diagnostic-test items,the Q-matrix should have more attributes than the number of the test items.