ABSTRACT
This chapter provides a historical and technical overview of statistical methods proposed to detect unusual response similarity and aberrant response patterns. It describes a simulation study investigating the performance of some of these methods under both nominal and dichotomous response outcomes. The chapter attempts to provide some insights to practitioners on the existing quantitative methods that have been developed to detect and combat answer copying/sharing through the use of both a simulation study and a demonstration of the use of these methods in a real test administration with flagged test takers and centers. The literature on statistical methods of detecting answer copying/sharing can be examined in two main categories: response similarity indices and person-fit indices. While some indices rely on developing and utilizing empirical null distributions, others use different distributional forms such as normal, binomial, Poisson, or compound binomial as a reference when computing the likelihood of observed agreement.
