An Iterative Method for Empirically-Based Q-Matrix Validation

Authors : Ragip Terzi, Jimmy De La Torre

Pages : 248-262

Doi:10.21449/ijate.407193

View : 18 | Download : 13

Publication Date : 2018-05-19

Article Type : Research

Abstract :In cognitive diagnosis modeling, the attributes required for each item are specified in the Q-matrix. The traditional way of constructing a Q-matrix based on expert opinion is inherently subjective, consequently resulting in serious validity concerns. The current study proposes a new validation method under the deterministic inputs, noisy "and” gate (DINA) model to empirically validate attribute specifications in the Q-matrix. In particular, an iterative procedure with a modified version of the sequential search algorithm is introduced. Simulation studies are conducted to compare the proposed method with existing parametric and nonparametric methods. Results show that the new method outperforms the other methods across the board. Finally, the method is applied to real data using fraction-subtraction data.
Keywords : Cognitive diagnosis model, Q-matrix validation, DINA, sequential search algorithm