The main purpose of the study is to examine the bias of the structural parameter when the multidimensionality was present. In this study, a bifactor model was used to generate the multidimensional data.
Questions and Comments:
1. In the table of data structure, the percentage of uncontaminatd correlations (PUC) was determined by the number of items, number of group factors and items per group factor. Hence, the effect of PUC is confunding with test length.
2. Since PUC can not be separated from the test length, OmegaH values may not affect by PUC (p. 11). Rather, it is actually affected by the test length.
3. The PUC does not directly indicate the size of multidimensionality. For structure 6, it seemed to be clear multidimensionality. Howeve, the PUC is the highest 0.94, which resulted in less bias (Figure 2).
4. The author examined whether the values of fit indices (RMSEA, CFI, SRMR) can be used to project bias by showing the correlation between the indices and EVC and omegaH. In addition, the indices were regressed on the bias of the structural coefficient. I think it is more promising to examine whether the indices is related to the factors that influence bias.