ABSTRACT
Measurement professionals and researchers have studied the problem of cheating detection for several years, with a marked increase in the number of articles in the past 10 to 15 years that describe answer-copying approaches for detecting potential test fraud (Belov ampentity Armstrong, 2009a; Maynes, 2005; Sotaridona ampentity Meijer, 2003; van der Linden ampentity Sotaridona, 2004, 2006; Wesolowsky, 2000; Wollack, 1997). The approaches vary between the authors, including the way the response probability model is computed; the requirement to designate a source and copier or not; and the use of incorrect matching responses or all responses. While incorrect matching responses provide the strongest evidence of potential test fraud, matching responses that are correct also provide evidence of potential test fraud.
