Detection of two-way outliers in multivariate data and application to cheating detection in educational tests

The paper proposes a new latent variable model for the simultaneous (two-way) detection of outlying individuals and items item-response-type data. proposed is synergy between factor binary responses continuous response times that captures normal item behaviour class items. A statistical decision framework developed under provides compound rules controlling local false discovery/nondiscovery rates outlier detection. Statistical inference carried out Bayesian which Markov chain Monte Carlo algorithm developed. method applied to cheating in educational tests, due leakage, using case study computer-based nonadaptive licensure assessment. performance evaluated by simulation studies.