1. It is stated that MANOVA will yield biased estimation. However, it is not clear that what kind of parameters are biased. And the paper does not provide the estimation for the three analyses (LR for a two-way interaction with restrict factorial invariance; LR test with a two-way interaction with a strong factorial invariance, and MANOVA);
2. I’m wondering what’s the purpose of conducting two LR tests? The paper did not provide clear explanations.
3. In IRT, the group difference can be studied using two approaches: (1) latent regression. But I’m not sure whether two-way interaction can be estimated using ConQuest; (2) plausible values approach.