There is increased use of Bayesian networks (BN) in educational assessment. In psychometrics, BN serves as a measurement model with high flexibility, suitable to model educational assessment data with a complex structure. BN is a novel psychometric approach and not all aspects of its application are well-known. The article aims to provide the systematization of BN model criticism methods in the field of educational assessment. The review revealed the diversity of model criticism methods and the shortages of specific research. The results demonstrate the state-of-the-art and help to navigate practitioners and researchers.