Parallel analysis is one of the common approaches of checking dimensionality of a latent variable. This article further uses simulation studies to confirm this is the most recommended methods for assessments. By comparing the performance of a number of approaches, Horn’s PA, PA-PAFA, and PA-MRFA, based on the Pearson correlation and Polychoric correlation-permutation, the authors arrived a conclusion that PA-MRFA is the best procedures. It is interesting to see that the authors obtained the sampling distribution of eigenvalues to complete their test. However, it has also left a question for researchers: how robust is this result when it is applied across other measurement settings, such as non-parametric models. Another thought is how this study replicated on some IRT-based models, such as a graded response model?