ABSTRACT
This chapter aims to explicate a method for evaluating group-level aberrance as potential evidence of cheating. This method focuses on unusual score gains as potential evidence of cheating. W. P. Skorupski and K. Egan have introduced a statistical method for detecting possible group-level cheating using a Bayesian Hierarchical Linear Model (BHLM). Bayesian procedures are used to estimate the parameters of a hierarchical linear model. The BHLM was fit to each replicated dataset, with estimation of a "growth aberrance" statistic used to evaluate potential cheating behavior. Groups simulated to display aberrant behavior were used for power analyses; nonaberrant groups were used to evaluate Type I error rates. Based on Markov Chain Monte Carlo (MCMC) convergence diagnostics, results indicate that the parameters of the hierarchical growth model converged to stable solutions. An MCMC algorithm produces random draws from the posterior distribution of each parameter being estimated.
