ABSTRACT
This conclusion presents some closing thoughts on the concepts covered in the preceding chapters of this book. The book begins with an examination of the status quo with respect to multiple response similarity measures and person-fit measures. It outlines a selection of person-fit and a machine-based learning approach to detecting aberrant response patterns. The book provides an excellent set of recommendations for practitioners to help them better understand how these indices will perform and should be interpreted in their testing programs. It focuses on M4, which is perhaps the most widely used and researched similarity index in practical settings. The book also provides an overview of issues and recommendations for testing programs as they further invest in up-to-date technology for their delivery platforms, particularly those coming out of the large collaborative efforts for state accountability systems. It presents the rationale behind Bayes' Theorem, and arguments for why this approach is superior in many ways to the more common frequentist approaches.
