A Note on the Invariance of the DINAL Model Parameters
by Jimmy de la Torre & Young0Sun Lee (2010)
Recent applications and works of cognitive diagnosis modeling show researchers’ increasing awareness and recognition of the potential benefits of this approach. However, for cognitive diagnosis models (CDMs) to receive wilder use, researchers and practitioners need to understand the basic properties of these models. In particular, it is important to understand how the parameters of these models are affected by changes in the characteristics of the underlying attribute distribution. In addition, it is also of value to document how the presence of noise in real data can skew these properties. The purpose of this article is that examined how the DINA model item parameter estimates vary as the characteristics of the underlying latent distribution change. The result shows that the invariance property of the DINA model parameters held across attribute distributions only to the extent that the model fits the data. It is worth noting that unlike traditional IRM, the invariance property of the DINA model is absolute in that the parameter estimates obtained using different calibration samples did not require any transformation for them to be comparable.