Exploring the Full-Information Bifactor Model in Vertical Scaling With Construct Shift
Ying Li and Robert W. Lissitz
To address the lack of attention to construct shift in IRT vertical scaling, the author proposed to use bifactor model for estimating the common dimension for all grades and the grade-specific dimension for each grade. For bifactor model, all the common items load only on the general factor, and grade-specific items load on corresponding grade-specific factors in addition to the general factor.
The simulation and empirical study were used for demonstration.
The results show that bifactor models were well recovered overall, though the grade-specific dimensions were not as well recovered as the general dimension.
Questions：
1 For the common items in the model which measures two group-specific ability/factor which is an violate for the one grade measure one ability assumption？
2 In the real situation, the group-specific ability are not fixed and usually difference from each other, how to deal with this situation？