Statistical Methods for Detecting Group-Based Test Fraud

• Activity Number: 453
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Government Statistics
• Author(s): Dennis Maynes*+
• Companies: Caveon Test Security
• Keywords: cheating ; erasure analysis ; answer-copying ; test fraud ; person fit analysis ; item response theory

Abstract

Several different statistical methods for detecting group-based test fraud have been proposed and used by practitioners. There has been very little confirmatory research on the efficacy of these methods. This presentation will provide a survey for methods that have been used by practitioners, where the methods have been used, the types of fraud that the methods intend to detect, and descriptions of utility of the methods. The methods include erasure analysis, similarity analysis (aka answer-copying detection), gain score analysis, aberrance or person fit analysis, and within-test score difference analysis. Underlying statistical theory will be presented for using these analyses to detect group-based test fraud. Methods for combining evidence from these statistical analyses will be presented for strengthening the inferences concerning test fraud. Finally, the state of current research in this area will be summarized in order to assist researchers as they contemplate working in this area.