ISSN 1671-3710
CN 11-4766/R

›› 2008, Vol. 16 ›› Issue (1): 175-183.

Previous Articles     Next Articles

Statistical Methods for Detecting Answer Copying

Han Dan;Guo Qingke;Wang Zhao;Chen Xuexia   

  1. Department of Psychology, Liaoning Normal University, Dalian 116029, Chin
  • Received:2007-11-06 Revised:1900-01-01 Online:2008-01-15 Published:2008-01-15
  • Contact: Guo Qingke

Abstract: Answer copying is the hardest form of cheating to detect. There are two types of statistical methods to detect it: answer-copying statistics (ACS) and person-fit statistics (PFS). In ACS the likelihood of two examinees to give same answer pattern is calculated, the copier is detected if the likelihood is rather small. PFS compares an observed item score pattern with a test model predicted, examinees who give aberrant item pattern are suspected to be the copier. Many other factors besides answer copying can give rise to inflated PFS. While ACS is especially presented to detect answer copying. Many researches were found that the detection rate of ACS was higher than PFS. ACS has been broadly used to detect answer copying in SAT, Licensure and certification examinations in America

Key words: answer copying, answer-copying statistics, person-fit statistics, Neural Network

CLC Number: