Exploring the Full-Information Bifactor Model in Vertical Scaling with Construct Shift
It’s assumed that item parameter of common item would not be affected by different groups (here is grade) for common item design. However, it is allowed common item with construct shift in the present study, which is the first question I want to ask. Due to existing of construct shift, the authors suggested that the bifactor model which can deal with such situation, and that general latent trait and group-specific trait play role in common items. That is, group-specific trait would be extracted if construct shift occurs. Second, it would be much better if the authors can provide some real example for helping readers to understand which kind of common item would change in item property under medical, educational, or psychological testing.